1.A tale of two languages

In his customary florid style, Leonard Palmer (1906–1984) called Vulgar Latin a “shimmering mirage” that becomes visible only when writers momentarily ignore the normative rules of their school learning: “There are, as it were, in the dead landscape of literary Latin, seismic areas where occasional eruptions reveal the intense subterranean activity which one day will make a new world of language” (1954: 149). This conjures up the image of a submerged variety of the language, visible only occasionally through errors in informal texts and inscriptions. Palmer underscores the variety of different colloquial styles, not all of which made it into the literary language. Some of these, like the colloquial language of the educated elite, could hardly be called ‘Vulgar Latin’; nonetheless, Palmer believes they all are somehow connected with the emergence of the Romance languages. Palmer’s characterization of Vulgar Latin may have rhetorical qualities, but the value of this notion for understanding the origin of the Romance languages is questionable.

In the present paper I shall first deal with the emergence of the idea of a variety called Vulgar Latin as a way to explain the relationship between the Latin of the literary texts and the Romance languages. Section 2 introduces two authors who approached this relationship from a different angle by focusing on the acquisition procss of the language in the provinces of the Roman empire, the Italian Celso Cittadini (1533–1627) and the Frenchman Pierre-Nicolas Bonamy (1694–1770), who credited the Roman soldiers with the transmission of the language. Section 3 discusses the role of the army in the process of Latinization. Section 4 looks at the chronology of the interaction between Roman conquerors and indigenous population, and Section 5 pinpoints the breaking off point, when the Romance languages began to be regarded as independent languages.

In 1979 Paul Lloyd remarked that “[i]n fact, there has been no progress at all in the definition of ‘Vulgar Latin’, but simply a constant repetition of previous definitions” (1979Lloyd, Paul M. 1979 “On the Definition of ‘Vulgar Latin’: The eternal return”. Neuphilologische Mitteilungen 80:2.110–122.: 113). He gave vent to his amazement that the confusion surrounding the term was commonly acknowledged by scholars, but did not lead to any initiative to discard it once and for all (1979: 120; see Adams 2013Adams, James N. 2013 Social Variation and the Latin Language. Cambridge: Cambridge University Press. : 5; Lüdtke 2016Lüdtke, Helmut 2016 Der Ursprung der romanischen Sprachen: Eine Geschichte der sprachlichen Kommunikation. 2nd ed. Kiel: Westensee-Verlag. (1st ed. 2009.): 149, 479; Eskhult 2018Eskhult, Josef 2018 “Vulgar Latin as an Emergent Concept in the Italian Renaissance (1435–1601): Its ancient and medieval prehistory and its emergence and development in Renaissance linguistic thought”. Journal of Latin Linguistics 17:2.191–230. : 192).

Lloyd’s warning about the risk of reifying ‘Vulgar Latin’ has not lost its urgency today; it is repeated by Nigel Vincent (2016Vincent, Nigel 2016 “Continuity and Change from Latin to Romance”. Early and Late Latin: Continuity or Change? ed. by James N. Adams & Nigel Vincent, 1–13. Cambridge: Cambridge University Press. : 7) with respect to the routine treatment of ‘Vulgar’ Latin as an intermediate language variety, serving as the ‘missing link’ between Classical Latin and the Romance languages. The introduction of this construct is commonly thought to go back to the establishment of Romance studies in Germany in the early nineteenth century (Swiggers 2001Swiggers, Pierre 2001 “Les débuts et l’évolution de la philologie romane au XIXe siècle, surtout en Allemagne”. History of the Language Sciences ed. by Sylvain Auroux, Konrad Koerner, Hans-Josef Niederehe & Kees Versteegh, II, 1272–1285. Berlin & New York: W. de Gruyter.). Right at the beginning of his comparative grammar of the Romance languages, Friedrich Diez (1794–1876) reminds his readers that these languages did not derive from Classical Latin:

Alle [romanische Sprachen] haben ihre erste und vornehmste Quelle in der lateinischen. Aber nicht aus dem classischen, dessen sich die Schriftsteller bedienten, flossen sie, sondern …, aus der römischen Volkssprache, welche neben dem classischen Latein im Gebrauch war.(Diez 1836Diez, Friedrich 1836 Grammatik der romanischen Sprachen, I. Bonn: E. Weber.: I, 1)

He mentions François Raynouard’s (1761–1836) proposal in his study on the troubadours (1821Raynouard, François Just Marie 1821 Grammaire comparée des langues de l’Europe latine dans leurs rapports avec la langue des troubadours. Paris: Firmin Didot.) to posit a common ancestor for the contemporary Romance languages, but does not follow the latter’s identification of this Proto-Romance with Old Provençal. Instead, he looks for the origin of the Romance languages in the colloquial language of the Romans (römische Volkssprache or Volksmundart).

Later German writers, such as Hugo Schuchardt (1842–1927) in his Vokalismus des Vulgärlateins (1866Schuchardt, Hugo 1866 Der Vokalismus des Vulgärlateins, I. Leipzig: B.G. Teubner.) and Wilhelm Meyer-Lübke (1861–1936) in his Grammatik der romanischen Sprachen (1890), did use the term Vulgar Latin (Vulgärlatein[isch]), although Meyer-Lübke restricted its use to the Latin spoken in areas outside the Romania, for instance in North Africa, as well as to those reconstructed forms that deviated from the regular correspondence with Latin (1890Meyer-Lübke, Wilhelm 1890 Grammatik der romanischen Sprachen, I. Leipzig: Fues’s Verlag.: I, 6). The latter use is clearly dependent on the paradigm of historical comparative linguistics; it was applied by him only reluctantly:“Vor dem Mißbrauch, wonach man dem Vulgärlatein alles in die Schuhe schickt, was man im Augenblick im Romanischen nicht zu erklären vermag, kann nicht genug gewarnt werden” (1890: I, 7).

This warning notwithstanding, throughout the nineteenth century the standard practice in treating the origin of the Romance languages included culling Vulgar Latin elements from the corpus of informal texts (Jorge 2018Jorge, Muriel 2018 Philologie, grammaire historique, histoire de la langue: Constructions disciplinaires et savoirs enseignés (1867–1923). Thèse de doctorat, Université Sorbonne Nouvelle–Paris 3.). Some preferred to call such elements collectively ‘Vulgar Latin’, others opted for different labels.

Latin authors had their own terminology to refer to non-literary elements in the language. When Cicero (106–43 BCE) or Quintilian (c. 35–c. 100 CE) use such terms as sermo plebeius or sermo vulgaris, they refer to casual, everyday speech (Adams 2013Adams, James N. 2013 Social Variation and the Latin Language. Cambridge: Cambridge University Press. : 12–22). This ‘vulgar Latin’ had nothing to do with the notion of Vulgar Latin as the Volkssprache from which the Romance languages were supposed to derive. It is more akin to its use in Dante’s (1265–1321) De vulgari eloquentia. With this term, Dante referred to the Italian vernacular language of his time (see Diez 1836Diez, Friedrich 1836 Grammatik der romanischen Sprachen, I. Bonn: E. Weber.: I, 76f.; Tavoni 1990Tavoni, Mirko 1990 “ ‘Ydioma tripharium’ (Dante, De vulgari eloquentia, I 8–9)”. History and Historiography of Linguistics: papers from the fourth International Conference on the History of the Language Sciences (ICHoLS IV), Trier, 24–28 August 1987 ed. by Hans-Josef Niederehe & Konrad Koerner, I, 233–247. Amsterdam & Philadelphia: John Benjamins. ; Gensini 2009Gensini, Stefano 2009 “Dante Alighieri”. Lexicon grammaticorum: A bio-bibliographical companion to the history of linguistics ed. by Harro Stammerjohann, I, 351–356. 2nd ed. Tübingen: Max Niemeyer.). He was convinced that the status of this vernacular could be raised by producing literary works, so that it would become as illustre as Latin itself (Holtus 1989Holtus, Günter 1989 “Das ‘vulgare illustre’ als Modell einer italienischen Kunstsprache: Standard, Substandard und Varietät in Dante Alighieri’s Traktat ‘De vulgari eloquentia’ (1305)”. Sprachlicher Substandard. II. Standard und Substandard in der Sprachgeschichte und in der Grammatik ed. by Günter Holtus & Edgar Radtke, 1–13. Tübingen: Niemeyer. ).

According to Dante (in Tavoni 2017Tavoni, Mirko ed. 2017 Dante Alighieri, De vulgari eloquentia. Milan: Mondadori.), the Romance vernaculars derived from one language (De vulgari eloquentia I, 8, 5), but, as Josef Eskhult (2018Eskhult, Josef 2018 “Vulgar Latin as an Emergent Concept in the Italian Renaissance (1435–1601): Its ancient and medieval prehistory and its emergence and development in Renaissance linguistic thought”. Journal of Latin Linguistics 17:2.191–230. : 200) clarifies, he did not believe that this language was Latin. In fact, he claimed it was the other way round: Latin derived from that same original Romance language, having been created by the Romans as a locutio secundaria. Dante postulated that, in addition to their spoken language, some language communities, among them the Greeks and the Romans (Schöntag 2017bSchöntag, Roger 2017b “Il dibattito intorno al volgare antico tra Leonardo Bruni e Flavio Biondo sullo sfondo della cognizione linguistica di Dante”. Forum Italicum 51:3.553–572. : 565f.), developed a secondary language which was called grammatica. This locutio secundaria was an invented, artificial language, intended to counter the variability of the spoken language that represented a threat to mutual understanding in the speech community.

The opposition between two locutiones, a spoken native language and an artificial learned language (which Dante calls gramatica) should not be interpreted in terms of diglossia or bilingualism, as is sometimes assumed (Eskhult 2018Eskhult, Josef 2018 “Vulgar Latin as an Emergent Concept in the Italian Renaissance (1435–1601): Its ancient and medieval prehistory and its emergence and development in Renaissance linguistic thought”. Journal of Latin Linguistics 17:2.191–230. : 199; Schöntag 2017bSchöntag, Roger 2017b “Il dibattito intorno al volgare antico tra Leonardo Bruni e Flavio Biondo sullo sfondo della cognizione linguistica di Dante”. Forum Italicum 51:3.553–572. : 559; Tavoni 2017Tavoni, Mirko ed. 2017 Dante Alighieri, De vulgari eloquentia. Milan: Mondadori.: 11). Dante believed that the written standard had been invented as a response to the growing variability in spoken language. The difference between the vernacular language and the standard language was formulated by him in terms of nature (secundum naturam) vs. convention (secundum placitum), or in terms of being learnt spontaneously by small children vs. being learnt at school at a more advanced age and used only by a literary elite. Dante does not call the two varieties linguae, but locutiones (Imbach & Rosier-Catach 2005Imbach, Ruedi & Irène Rosier-Catach 2005 “De l’un au multiple, du multiple à l’un: Une clef d’interprétation pour le De vulgari eloquentia ”. Mélanges de l’École française de Rome. Moyen Âge 117:2.509–529.), a term typical of logic and philosophy, rather than linguistics.11.It is beyond the scope of the present article to investigate the sources of Dante’s distinction between two locutiones. He was certainly familiar with the linguistic theories of the Modistae (Alessio 2015Alessio, Gian Carlo 2015 “La grammatica speculativa e Dante”. Gian Carlo Alessio, Lucidissima dictandi peritia: Studi di grammatica e retorica medievale ed. by Filippo Bognini, 127–144. Venice: Edizioni Ca’ Foscari.: 136), and there are parallels between his distinction and the one set up by Robert of Kilwardby (c. 1215–1279), according to whom the science of grammar was needed to re-establish order in natural speech (Alessio 2015Alessio, Gian Carlo 2015 “La grammatica speculativa e Dante”. Gian Carlo Alessio, Lucidissima dictandi peritia: Studi di grammatica e retorica medievale ed. by Filippo Bognini, 127–144. Venice: Edizioni Ca’ Foscari.: 138). Lo Piparo (1986)Lo Piparo, Franco 1986 “Sign and Grammar in Dante: A non-Modistic language theory”. The History of Linguistics in Italy ed. by Paolo Ramat, Hans-Josef Niederehe & Konrad Koerner, 1–22. Amsterdam & Philadelphia: J. Benjamins. acknowledges Dante’s familiarity with the Modists’ writings, but stresses the difference in the agenda underlying his gramatica and the considerations of Modistic grammar. At any rate, it is clear that his ideas are quite different from the preoccupations of modern sociolinguistics.

At first, Dante’s ideas did not play a significant role in the debates in Italy about the nature of Latin (Marazzini 2018Marazzini, Claudio 2018 Breve storia della questione della lingua. Rome: Carocci.: 16–17), but the political context of the Italian Quattrocento brought the issue of the language spoken by the Romans in Antiquity to the fore (Raffarin 2015Raffarin, Anne ed. 2015 Flavio Biondo, Leonardo Bruni, Poggio Bracciolini, Lorenzo Valla: Débats humanistes sur la langue parlée dans l’Antiquité. Textes édités, traduits, présentés, annotés et commentés. Paris: Les Belles Lettres.: 29–31), leading to a renewed interest in his treatise. The two main positions in what came to be called the questione della lingua (Schöntag 2017aSchöntag, Roger 2017a “Das Verständnis von Vulgärlatein in der Frühen Neuzeit vor dem Hintergrund der questione della lingua: Eine Untersuchung zur Begriffsgeschichte im Rahmen einer varietätenlinguistischen Verortung”. Synchronie und Diachronie (= Bavarian Working Papers in Linguistics, 6) ed. by Barbara Sonnenhauser, Caroline Trautmann, Daniel Holl & Patrizia Noel Aziz Hanna, 111–129. Munich: Ibykos Verlag.) were represented by Flavio Biondo (1392–1463) and Leonardo Bruni (1370–1444). According to Biondo (2015 [1435])Biondo, Flavio 2015 [1435] De verbis romanae locutionis Biondi ad Leonardum Arretinum. Flavio Biondo, Leonardo Bruni, Poggio Bracciolini, Lorenzo Valla: Débats humanistes sur la langue parlée dans l’Antiquité. Textes édités, traduits, présentés, annotés et commentés ed. by Anne Raffarin, 123–169. Paris: Les Belles Lettres., everyone in Ancient Rome had spoken Latin; variation between the speakers was based on a difference in style and had not led to a split between separate languages. In Biondo’s view, contemporary Italian was basically Latin, albeit a form of the language that had become corrupted during the fifth-century Barbarian invasions (Schöntag 2017aSchöntag, Roger 2017a “Das Verständnis von Vulgärlatein in der Frühen Neuzeit vor dem Hintergrund der questione della lingua: Eine Untersuchung zur Begriffsgeschichte im Rahmen einer varietätenlinguistischen Verortung”. Synchronie und Diachronie (= Bavarian Working Papers in Linguistics, 6) ed. by Barbara Sonnenhauser, Caroline Trautmann, Daniel Holl & Patrizia Noel Aziz Hanna, 111–129. Munich: Ibykos Verlag.: 117f.; Marcellino 2019Marcellino, Giuseppe 2019 “Latin and the Vernacular in Biondo Flavio’s Thought and Works”. Neo-Latin and the Vernaculars: Bilingual interactions in the Early Modern Period ed. by Alexander Winkler & Florian Schaffenrath, 11–35. Leiden: E.J. Brill.).

Bruni (2015 [1435]Biondo, Flavio 2015 [1435] De verbis romanae locutionis Biondi ad Leonardum Arretinum. Flavio Biondo, Leonardo Bruni, Poggio Bracciolini, Lorenzo Valla: Débats humanistes sur la langue parlée dans l’Antiquité. Textes édités, traduits, présentés, annotés et commentés ed. by Anne Raffarin, 123–169. Paris: Les Belles Lettres.), on the other hand, claimed that a vulgar language had always coexisted with the Classical language. According to him, Italian was just the same impure form of Latin that had always existed (Schöntag 2017aSchöntag, Roger 2017a “Das Verständnis von Vulgärlatein in der Frühen Neuzeit vor dem Hintergrund der questione della lingua: Eine Untersuchung zur Begriffsgeschichte im Rahmen einer varietätenlinguistischen Verortung”. Synchronie und Diachronie (= Bavarian Working Papers in Linguistics, 6) ed. by Barbara Sonnenhauser, Caroline Trautmann, Daniel Holl & Patrizia Noel Aziz Hanna, 111–129. Munich: Ibykos Verlag.: 117). Therefore, he claimed, all writers had to do was revert to the use of pure Latin. According to Eskhult (2018Eskhult, Josef 2018 “Vulgar Latin as an Emergent Concept in the Italian Renaissance (1435–1601): Its ancient and medieval prehistory and its emergence and development in Renaissance linguistic thought”. Journal of Latin Linguistics 17:2.191–230. : 202), Bruni’s ideas came close to Dante’s view of the language situation in Ancient Rome when he called the two distinct varieties of Latin prima and secundaria: the primary variety was the language people learnt from their parents, which they spoke on the basis of natural instinct, whereas the lingua secundaria was the grammatical language learnt at school. Unlike Dante, however, Bruni did not believe that the vernacular language could be refined by developing a secundary variety with grammatical rules, but thought that the best solution was to avoid the popular variety altogether, employing only the grammatical language. In this respect, it was Biondo whose ideas were more in line with Dante’s view that the difference between the two locutiones resided in the lack of latinitas of the forma vulgaris. Biondo’s solution consisted in developing grammatical rules for the volgare.

2.A new idea

None of the arguments advanced for either position in the debate about the language situation in Ancient Rome made any reference to the actual process of language acquisition that had led to the spreading of Latin in the provinces. Most of the participants in the debate, Biondo and Bruni included, took for granted that the inhabitants of the provinces had taken over the new language wholesale, whether this was the pure Classical language or an impure version of it. Very few people thought differently. One of them was Celso Cittadini (1533–1627), who in 1601 published his Trattato della vera origine, e del processo, e nome della nostra lingua scritto in volgar sanese (Swiggers 2014Swiggers, Pierre 2014 “Les études linguistiques romanes des origines jusqu’au début du XIXe siècle: Les ‘prémices’ de la romanistique”. Manuel des langues romanes ed. by André Klump, Johannes Kramer & Aline Willems, 13–42. Berlin and New York: W. de Gruyter. : 19; Schöntag 2017aSchöntag, Roger 2017a “Das Verständnis von Vulgärlatein in der Frühen Neuzeit vor dem Hintergrund der questione della lingua: Eine Untersuchung zur Begriffsgeschichte im Rahmen einer varietätenlinguistischen Verortung”. Synchronie und Diachronie (= Bavarian Working Papers in Linguistics, 6) ed. by Barbara Sonnenhauser, Caroline Trautmann, Daniel Holl & Patrizia Noel Aziz Hanna, 111–129. Munich: Ibykos Verlag.: 124–126; Cittadini 2019 [1601]Cittadini, Celso 2019 [1601] Trattato della vera origine e del processo, e nome della nostra lingua, scritto in volgar sanese (Venezia Ciotti). Edited by Pierluigi Ortolano. Florence: Franco Cesati. [Originally published in Venice: Ciotti 1601.]), in which he claimed that Latin had always been divided into “due manieri di lingua, l’una pura latina … e l’altra mescolata di barbarismi e di falsilatini del vulgo e de’ cittadini e de’ contadini e de’ forestieri idioti e senza lettere” (Cittadini 2019 [1601]Cittadini, Celso 2019 [1601] Trattato della vera origine e del processo, e nome della nostra lingua, scritto in volgar sanese (Venezia Ciotti). Edited by Pierluigi Ortolano. Florence: Franco Cesati. [Originally published in Venice: Ciotti 1601.]: 123f.), rejecting the idea that Italian was the product of a process of language corruption during the Barbarian invasions. Eskhult (2018Eskhult, Josef 2018 “Vulgar Latin as an Emergent Concept in the Italian Renaissance (1435–1601): Its ancient and medieval prehistory and its emergence and development in Renaissance linguistic thought”. Journal of Latin Linguistics 17:2.191–230. : 221) interprets Cittadini’s ideas as an early conceptualization of diglossia. Yet, this is not in accordance with Cittadini’s description: his lingua antiqua del vulgo is the language of the vulgus, but this does not make it identical with the Low variety in Ferguson’s (1959)Ferguson, Charles A. 1959 “Diglossia”. Word 15.325–40. notion of diglossia, which functions as the everyday language for all members of the speech community, not only for the lower classes. To some extent, Cittadini agrees with Bruni that the two ways of speaking had always coexisted. Unlike Bruni, however, he links this with the acquisition of the language in the provinces, holding the Roman soldiers responsible for the transmission of a mutilated form of Latin:

Oltracciò i soldati delle legioni che per la maggior parte erano delle provincie e per conseguenza di varii e diversi linguaggi ed anco gli stessi nati in Roma, ma stati longamente fuora alle guerre o nelle guarnigioni o per altro in diverse parti del mondo, quando venivano o tornavano a Roma, con la conversation loro e uso di nuove lingue, corrompevano la purità della lingua latina in coloro che l’havevano, parlando essi mezzo barbaramente, cioè volgarmante [sic].(Cittadini 2019 [1601]Cittadini, Celso 2019 [1601] Trattato della vera origine e del processo, e nome della nostra lingua, scritto in volgar sanese (Venezia Ciotti). Edited by Pierluigi Ortolano. Florence: Franco Cesati. [Originally published in Venice: Ciotti 1601.]: 128)

Cittadini did not go so far as to speculate about the language type resulting from this process, but just stated that it was the soldiers’ language or languages that ended up becoming the native language of the new speakers.

A more elaborate view of the process of language transmission is found almost a century later in the Mémoires presented by Pierre-Nicolas Bonamy (1694–1770) to the Académie Royale des Inscriptions et Belles-Lettres (Albrecht 1975Albrecht, Jörn ed. 1975 Pierre-Nicolas Bonamy: Vier Abhandlungen zum Vulgärlatein und zur Frühgeschichte des Französischen, ausgewählt, eingeleitet und kommentiert. Tübingen: Gunter Narr.; Swiggers 2014Swiggers, Pierre 2014 “Les études linguistiques romanes des origines jusqu’au début du XIXe siècle: Les ‘prémices’ de la romanistique”. Manuel des langues romanes ed. by André Klump, Johannes Kramer & Aline Willems, 13–42. Berlin and New York: W. de Gruyter. : 27–28; Jorge 2018Jorge, Muriel 2018 Philologie, grammaire historique, histoire de la langue: Constructions disciplinaires et savoirs enseignés (1867–1923). Thèse de doctorat, Université Sorbonne Nouvelle–Paris 3.: 392). Bonamy states that it is obvious that French, Spanish, and Italian derived from the “langue vulgaire des provinces”, and not from Classical Latin (1756: 597). Some of his peers were reluctant to accept the relationship between French and Latin and preferred to link French with the Celtic language that had been spoken in France before the Roman conquests. Bonamy believes that this mistaken belief stemmed from the fact that they compared French with Classical Latin (1756: 607), without taking into account the context of language acquisition:

Comment, après tout, le simple peuple, parmi les Gaulois, avoit-il appris la langue Latine? ce n’étoit assurément pas dans les Académies dont j’ai parlé, ni dans les livres écrits en cette langue; ce ne pouvoit être qu’en l’entendant prononcer aux Romains, soldats, marchands, artisans, esclaves, qui n’avoient pas plus fréquenté les écoles d’Italie que les Gaulois celles des Gaules.(Bonamy 1756Bonamy, Pierre-Nicolas 1756 Mémoire sur l’introduction de la langue latine dans les Gaules, sous la domination des Romains. Mémoires de Littérature tirés des registres de l’Académie Royale des Inscriptions et Belles-Lettres XXIV, 582–602. Paris: L’Imprimerie Royale. (Repr. in Pierre-Nicolas Bonamy: Vier Abhandlungen zum Vulgärlatein und zur Frühgeschichte des Französischen, ausgewählt, eingeleitet und kommentiert ed. by Jörn Albrecht, 1–22. Tübingen: Gunter Narr 1975.): 594)

Bonamy sees the role of the soldiers as providing the model for the new learners who learnt the language “par la fréquentation avec les Romains de toutes sortes de conditions” (1756: 641). He did not consider the soldiers’ language to be a corrupted form of Latin (Albrecht 1975Albrecht, Jörn ed. 1975 Pierre-Nicolas Bonamy: Vier Abhandlungen zum Vulgärlatein und zur Frühgeschichte des Französischen, ausgewählt, eingeleitet und kommentiert. Tübingen: Gunter Narr.: 27). For him, the Latin spoken by those who came to the provinces as soldiers or traders was simply the popular variety of Latin, representing the natural development of the language, while Classical Latin was a construct, mirroring Dante’s lingua secundaria.

Bonamy’s explanation of the difference between Classical Latin and the Romance languages has not been taken over by the communis opinio in Romance studies. Modern handbooks of the history of the Latin language still routinely include a chapter on Vulgar Latin, treating it as an intermediate variety between Latin and the Romance languages, without concerning themselves with the actual process of transmission of the language, but taking for granted that it was this variety that was adopted intact by the indigenous population.22.For an older status quaestionis see Mohrmann (1951)Mohrmann, Christine 1951 “Les formes du latin dit ‘vulgaire’: Essai de chronologie et de systématisation de l’époque augustinéenne aux langues romanes”. Actes du Premier Congrès d’études classiques, 207–220. Paris: Klincksieck., who focuses on the differentiation of Latin as the result of a number of sociolinguistic circumstances, including substratal influence and the impact of Christianization, and Tovar (1964)Tovar, Antonio 1964 “A Research Report on Vulgar Latin and its Local Variations”. Kratylos 9.113–134., who warns against the concept of a unitary Vulgar Latin and draws attention to factors of differentiation. Kiesler (2018)Kiesler, Reinhard 2018 Einführung in die Problematik des Vulgärlateins ed. by Volkert Noll. 2nd ed. Berlin and Boston: de Gruyter. is a more recent summary of the different positions and problems. Väänänen (1981Väänänen, Veikko 1981 Introduction au latin vulgaire. 3rd ed. Paris: Klincksieck. (Repr. 2012; 1st ed. 1963.): 4–6) describes the modern discussions on the status of Vulgar Latin as a tension between two conceptions. On the one hand, there is the tendency to regard Classical Latin and Vulgar Latin as basically one language, where the latter is the popular or colloquial register of the former. Grandgent’s Introduction to Vulgar Latin (1907Grandgent, Charles Hall 1907 An Introduction to Vulgar Latin. Boston: D.C. Heath.: 3), for instance, gives the following definition: “What we call Vulgar Latin is the speech of the middle classes, as it grew out of early Classic Latin”. Grandgent adds that this Vulgar Latin was different from the language of the cultural elite, but also from that of the rustic speakers in the countryside, as well as from that of the lowest layers of society. As sources for the difference between Vulgar and Classical Latin he employs observations by the Latin grammarians, as well as the informal texts that have since become the canon for Vulgar Latin. He believes that these materials provide glimpses of a spoken variety of Latin that was responsible for the emergence of the ‘new’ Romance languages, and concludes from this that spoken Latin died out in the sixth or seventh century CE, when the Romance languages replaced it. According to Grandgent, those differences between the Romance languages and Classical Latin that cannot be documented in the ‘Vulgar Latin’ texts and that do not manifest themselves until after the emergence of the Romance languages, had always been there as submerged variants, which is basically what Bruni had proposed in the fifteenth century.

The alternative explanation posits a gradual change from Latin to the Romance languages. Eugenio Coseriu (1921–2002) suggests that at first Classical Latin and the colloquial language did not diverge very much, but at a later stage the developing colloquial language began to deviate from the codified Classical language (Coseriu 1954Coseriu, Eugenio 1954 “El llamado « latín vulgar » y las primeras diferenciaciones romances: Breve introducción a la lingüística románica”. Montevideo: Universidad de la República.). In essence, this is the same scenario as the one proposed by James Adams (2013Adams, James N. 2013 Social Variation and the Latin Language. Cambridge: Cambridge University Press. : 234). He does not believe in submerged varieties, but allows for variation in Latin across social layers, opposing the kind of reasoning that sees in isolated instances of a phenomenon the forerunner of a development that did not take place until much later (see also Clackson 2016aClackson, James 2016a “Latin as a Source for the Romance Languages”. The Oxford Guide to the Romance Languages ed. by Adam Ledgeway & Martin Maiden, 3–13. Oxford: Oxford University Press. : 6). In his view, Latin and the vernaculars grew apart gradually until the vernaculars were recognized as languages in their own right. This view bears more resemblance to Biondo’s position in the controversy.

In both scenarios, the image presented is that of Romans handing down their language as a wholesale package to the provincial populace. At first, the learners may have had to struggle in adapting to their new language, but, just like they accommodated to Roman culture and law, they ended up by taking over the language imposed on them. The next generations then became native speakers of Latin, perhaps with some slight modifications. Such approaches do not leave much room for a discussion of the acquisition process itself. It is generally accepted that accommodation to Roman culture and law was a process of negotiation between indigenous customs and imperial standards, which led to considerable regional differences. Yet, with respect to the acquisition of the language, scant attention is given to the processes involved. None of the nine definitions of Vulgar Latin listed in Reinhard Kiesler’s introduction (2018Kiesler, Reinhard 2018 Einführung in die Problematik des Vulgärlateins ed. by Volkert Noll. 2nd ed. Berlin and Boston: de Gruyter. : 10–12), for instance, makes any reference to the acquisition of Latin by second language learners. What these definitions have in common is the emphasis on Vulgar Latin as a spoken non-standard variety of Latin, either as a variety that has always been there, or as a form of Late Latin, for instance in the definition proposed by Coseriu (1954)Coseriu, Eugenio 1954 “El llamado « latín vulgar » y las primeras diferenciaciones romances: Breve introducción a la lingüística románica”. Montevideo: Universidad de la República.. For Veikko Väänänen (1905–1997) Vulgar Latin is the traditional (though not very appropriate) term for the successive stages of the language from the first century BCE till the first written testimonies of the Romance languages (Väänänen 1981Väänänen, Veikko 1981 Introduction au latin vulgaire. 3rd ed. Paris: Klincksieck. (Repr. 2012; 1st ed. 1963.): 4). But in all scenarios the Latin language is transmitted intact by native speakers to the inhabitants of the provinces.33.Noteworthy exceptions are Maximilian Křepinský (1875–1971), who refers to the time “lorsque les indigènes dans les provinces ont essayé de parler la langue de leurs vainqueurs” (1958: 1), which puts the responsibility for the changes squarely on the new learners; and Theodoro Maurer (1906–1979), who speaks of “uma população mista, que em grande parte aprendera o latim imperfeitamente por uma rapida assimilação” (Maurer 1962Maurer, Theodoro Henrique 1962 O problema do latim vulgar. Rio de Janeiro: Livraria Acadêmica.: 63).

In this respect, nothing much has changed between the questione della lingua debates and modern theories about the origin of the Romance languages. Michel Banniard’s (2011Banniard, Michel 2011 “Du latin tardif (IIIe-VIIe siècle) au protofrançais (VIIIe siècle): Vers un nouveau paradigme”. Diachroniques. Revue de linguistique française diachronique 1.39–58.: 42) assertion that “[a]bsolument rien n’empêchait tous les habitants de l’empire de devenir des locuteurs latinophones au même titre que leurs conquérants” reflects the same general idea about language acquisition that Biondo had formulated, implying that after the language shift had been completed, everyone spoke a language unaffected by any changes during the acquisition process. Along the same lines, Helmut Lüdtke (2016Lüdtke, Helmut 2016 Der Ursprung der romanischen Sprachen: Eine Geschichte der sprachlichen Kommunikation. 2nd ed. Kiel: Westensee-Verlag. (1st ed. 2009.): 401f.) maintains that younger people from the higher classes in the city were more adept at learning Latin than older people from the lower classes in the countryside and boldly claims that “[d]er Sprachwechsel war abgeschlossen, als die letzten älteren, auf dem Lande ansässigen Unterschichtssprecher der betreffenden vorrömischen Sprache gestorben waren”. This begs the question of the model these people followed in learning the language.

3.From whom did they learn Latin?

Is there any substance to the suggestions made by Cittadini and, after him, by Bonamy about the spreading of Latin through the intermediary of the Roman soldiers? Within a relatively short period of time the Roman Empire expanded to include vast areas in Europe, North Africa and the Levant. The conquests involved a swift military deployment, followed by the stationing of relatively small garrisons across a vast region amidst numerically superior masses of speakers of other languages. These small clusters of military presence served as centers of attraction for the subdued populations and as sources from which culture, customs, religion, and language could spread. In the western parts of the empire, once Roman administration and law had been introduced, large numbers of people adopted the language of the Romans as a second language. They did so as a matter of convenience or because of the prestige knowledge of the imperial language brought with it, not because learning the language was imposed on them. The language policy of the Roman administration did not involve compulsory learning of Latin (Rochette 2011Rochette, Bruno 2011 “The Language Policies in the Roman Republic and Empire”. A Companion to the Latin Language ed. by James Clackson, 549–563. Oxford: Blackwell. ). The farther away from the center of the empire, the scarcer the formal institutions for learning. Conversely, a province close to Rome, like Gaul, offered more opportunities for learning than other regions did, and these survived for a longer period of time, even after the fall of the empire (John 2022John, Alison 2022 “Cultural Memory and Classical Education in Late Antique Gaul”. Making and Unmaking Ancient Memory ed. by Rajiv K. Bhola & Martine De Marre, 256–272. New York & London: Routledge. ). But even there the common people did not have many chances to profit from the educational network, although they were no doubt exposed much more to contact with native speakers through the commercial channels.

For the vast majority of the population acquisition of the imperial language took place without monitoring, with only minimal participation of native speakers as a model. Most of the learners lacked any literacy skills (Becker 2014Becker, Lidia 2014 “La protohistoire médiévale des langues romanes”. Manuel des langues romanes ed. by André Klump, Johannes Kramer & Aline Willems, 261–286. Berlin & New York: W. de Gruyter. : 274, referring to Goetz 2003Goetz, Hans-Werner 2003 Europa im frühen Mittelalter: 500–1050. Stuttgart: Ulmer. : 33), and even when they received a modicum of schooling, they were far from proficient when trying their hand at writing.44.Based on the epigraphical evidence it is sometimes assumed that a sizable percentage of the population of the provinces were literate. Curchin (2004Curchin, Leonard A. 2004 The Romanization of Central Spain: Complexity, Diversity and Change in a Provincial Hinterland. London & New York: Routledge. : 206–214), for instance, claims that many more people in Central Spain were literate than is commonly estimated because of the number of inscriptions, but this is rather unlikely. After all, having a monument with an inscription erected in one’s honor does not mean that one can actually read the inscription. It just means that one is following the Roman custom without necessarily being proficient in the language of the inscription. A more balanced judgment is given by Nedeljković (2015Nedeljković, Vojin 2015 “Castles Made of Sand? Balkan Latin from Petar Skok to J.N. Adams”. The Danubian Lands between the Black, Aegean and Adriatic Seas (7th century BC – 10th century AD): Proceedings of the Fifth International Congress on Black Sea Antiquities (Belgrade 17–21 September 2013) ed. by Gocha R. Tsetskhladze, Alexandru Avram & James Hargrave, 323–328. Oxford: Archaeopress. : 325), who speaks of the “notoriously meagre ‘epigraphic layer’ of the population”. Rutgers (1995Rutgers, Leonard Victor 1995 The Jews in Late Ancient Rome: Evidence of Cultural Interaction in the Roman Diaspora. Leiden: E.J. Brill. (Repr. 2000.) : 189) appeals to common sense when he observes that “[i]nscriptions inform, in the first place, about the level of education of the person who carved these inscriptions. They inform only indirectly, however, about the level of education of the person ordering an inscription”. There was no doubt a provincial elite, consisting of wealthy people who could afford the best education for their children, but they were a tiny minority, for whom Roman culture and the Latin language became a new identity marker.55.For the role assigned to the provincial elites in the grand scheme of Roman expansion, see Millett (1990Millett, Martin 1990 “Romanization: Historical issues and archaeological interpretation”. The Early Roman Empire in the West ed. by Thomas Blagg & Martin Millett, 35–41. Oxford: Oxbow Books., especially 38). For the survival of the ‘Roman’ identity after the fall of the empire, in particular in the Gallo-Roman world, see Mathisen (2018)Mathisen, Ralph W. 2018 “ ‘Roman’ Identity in Late Antiquity, with Special Attention to Gaul”. Transformations of Romanness in the Early Middle Ages ed. by Walter Pohl, 255–274. Berlin: de Gruyter. ; John (2022)John, Alison 2022 “Cultural Memory and Classical Education in Late Antique Gaul”. Making and Unmaking Ancient Memory ed. by Rajiv K. Bhola & Martine De Marre, 256–272. New York & London: Routledge. . For the rest of the population, it was a prestigious model, to be emulated, but out of reach.

The model the Roman conquerors provided for language learning was not a standardized form of the language, frequently not even a native variety, since many Roman soldiers were new speakers themselves.66.Among the legionaries there may have been relatively more individuals who had attained a certain level of literacy, especially among the commanding officers (Speidel 1995Speidel, Michael Alexander 1995 “Das römische Heer als Kulturträger”. La politique édilitaire dans les provinces de l’Empire romain. IIème-IVème siècles après J.-C.: Actes du IIe Colloque roumano-suisse (Berne, 12–19 septembre 1993) ed. by Regula Frei-Stolba & Heinz E. Herzig, 187–209. Bern: P. Lang. (Repr., Heer und Herrschaft im römischen Reich der Hohen Kaiserzeit, ed. by Michael Alexander Speidel, 515–544. Stuttgart: F. Steiner 2009.)), but these were not necessarily the ones in frequent contact with the local population. In many parts of the empire, the Roman garrisons consisted of legionaries who had been recruited elsewhere (Haynes 1999Haynes, Ian 1999 “Military Service and Cultural Identity in the Auxilia ”. The Roman Army as a Community, including papers of a conference held at Birkbeck College, University of London on 11–12 January, 1997 ed. by Adrian Goldsworthy & Ian Haynes, 165–174. (= Journal of Roman Archaeology, supplementary series 34.) Portsmouth, RI: Journal of Roman Archaeology.: 166). There was no consistent policy to station auxiliaries as far away as possible from the place where they had been recruited, but there were often practical reasons for posting them from one province to another (Haynes 2013Haynes, Ian 2013 Blood of the Provinces: The Roman Auxilia and the Making of Provincial Society from Augustus to the Severans. Oxford: Oxford University Press. : 121f.). Besides, the regiments were in constant need of replenishing, so that most of them ended up being multiethnic anyway. Actually, this was likely to be regarded as less risky than ethnical homogeneity when it came to maintaining discipline (Haynes 2013Haynes, Ian 2013 Blood of the Provinces: The Roman Auxilia and the Making of Provincial Society from Augustus to the Severans. Oxford: Oxford University Press. : 206).

Adams (2013Adams, James N. 2013 Social Variation and the Latin Language. Cambridge: Cambridge University Press. : 856) treats the language of soldiers as a stylistic variety used by a particular social class. He cites the example of a decurion’s letter to his commanding officer in which the 3rd pers. pl. ending -unt is used with verbs of the -ēre conjugation, as in habunt, debunt, instead of the standard forms habent, debent. Adams adduces this as evidence showing that not all non-standard variants start in the lowest social classes. One could, however, also reason the other way round: if a decurion could use such forms in a letter to his superior officer, he is bound to have used forms that were far less standard in addressing non-native speakers, whether civilians or recruits. Even when the soldiers were native speakers, they must have resorted to a variety of foreigner-directed speech in their contacts with the indigenous population. This modified and reduced form of speech constituted the model emulated by the new learners of Latin. The main question is therefore which forms of Latin were selected when addressing new learners, whether Greek slaves in Rome, Gaulish peasants, German prisoners of war, or British traders.

The army’s impact on Latinization in the provinces extended beyond the immediate contact with legionaries. It is commonly assumed that after their active service veterans founded provincial centers from where Roman culture spread. Numerically, however, the presence of veterans cannot have been very significant. Haynes (2013Haynes, Ian 2013 Blood of the Provinces: The Roman Auxilia and the Making of Provincial Society from Augustus to the Severans. Oxford: Oxford University Press. : 341) calculates for the first/second centuries an average total number of veterans from among both the auxiliaries and the legionaries of 82,000 to 120,000, which amounts to approximately 0.2 to 0.3 percent of the total provincial population. This figure may be compared with Woolf’s (1998Woolf, Greg 1998 Becoming Roman: The Origins of Provincial Civilization in Gaul. Cambridge: Cambridge University Press. : 138) estimate of the total number of inhabitants of the three Gallias at twelve million, of which at most ten percent lived in urban centers, the rest in the countryside where there were even less opportunities of meeting with Roman soldiers or traders.77.For other estimates of the numerical strength of the provincial population at the time of the Roman conquests see Kiesler (2018Kiesler, Reinhard 2018 Einführung in die Problematik des Vulgärlateins ed. by Volkert Noll. 2nd ed. Berlin and Boston: de Gruyter. : 28–31), who quotes 10 million for Gallia and 3–5 million for Hispania; Italia had about 6–8 million inhabitants at the time of Augustus. Gulyás (2019Gulyás, Adrienn 2019 “How Do New Languages Arise? A comparison of romanization and gallicization”. Acta Antiqua Academiae Scientiarum Hungaricae 59.9–18. : 13) claims a maximum of 500,000 Romans in Gallia on a total of 10 million local inhabitants. Haynes (2013Haynes, Ian 2013 Blood of the Provinces: The Roman Auxilia and the Making of Provincial Society from Augustus to the Severans. Oxford: Oxford University Press. : 339–367) is therefore right in being somewhat skeptical about the actual impact of veterans in provincial society. He remarks that in many cases veterans tended to settle in the neighborhood of military establishments, or even in the extramural settlements outside the garrison. Veterans did receive land grants, but it is doubtful that after a long period of active military duty they would wish to take up a farmer’s life. In fact, “[a]uxiliary veterans seem to capitalize on what is there rather than create something afresh” (Haynes 2013Haynes, Ian 2013 Blood of the Provinces: The Roman Auxilia and the Making of Provincial Society from Augustus to the Severans. Oxford: Oxford University Press. : 366f.).

4.And when did they learn it?

The argument of gradual change in the development from Latin to the Romance languages is seemingly supported by the evidence of written texts. Yet, this evidence must be used with caution. Writing always imposes restrictions, and anyone using this medium has to adapt to the conventions of the written standard. Errors encountered in texts, whether inscriptions or papyri or manuscripts, cannot be adduced as direct evidence of the contemporary spoken language.88.This even applies to the matter of language choice in the community. Rutgers (1995Rutgers, Leonard Victor 1995 The Jews in Late Ancient Rome: Evidence of Cultural Interaction in the Roman Diaspora. Leiden: E.J. Brill. (Repr. 2000.) : 176–209) shows that the Jewish community in Rome continued to use Greek as the language of funerary inscriptions to a much larger degree than the Christian community. But this does not mean that the Jewish community was less Latinized: after all, even in the inscriptions that were written in Greek, Latinized names were used frequently. As long as formal education is a prerogative of the elite, errors occur only infrequently in writing, but as the average level of schooling grows, the number of semi-literate people increases as well, leading to an increased frequency of errors. Written texts are therefore always suspect as evidence of the progression of linguistic changes: what they represent is the fluctuating influence of the standard language (Versteegh 2002Versteegh, Kees 2002 “Dead or Alive: The status of the standard language”. Bilingualism in Ancient Society: Language Contact and the Written Word ed. by James N. Adams, Mark Janse & Simon Swain, 52–74. Oxford: Oxford University Press. ).99.The distinction is complicated by the fact that Classical writers may also use colloquialisms to satyrize the speech of certain people, as in Petronius’ Cena Trimalchionis, or to paint a couleur locale, as in Plautus’ comedies, or for the more mundane reason of lack of competence in written Latin, as in Egeria’s account of her travel to the Holy Land, or to emphasize their bond with the common believers, as in the case of Christian authors, or just simply for the fun of being vulgar, as in some of the graffiti in Pompeii (Clackson 2016bClackson, James 2016b “The language of a Pompeian Tavern: Submerged Latin?”. Early and Late Latin: Continuity or Change? ed. by James N. Adams & Nigel Vincent, 70–86. Cambridge: Cambridge University Press. ). In all of these examples the authors are native speakers.

The confusion in the use of declensional endings is a case in point. Béla Adamik (2019)Adamik, Béla 2019 “Transformation of the Case System in A. L. as Evidenced in Inscriptions”. Acta Classica Universitatis Scientiarum Debreceniensis 55.13–36. discusses mistakes in inscriptions from Latin Africa in connection with József Herman’s (1924–2005) thesis about the collapse of the case system in African Latin (Herman 2000 [1967]Herman, József 2000 [1967] Vulgar Latin. English transl. by Roger Wright. University Park, Penn.: The Pennsylvania State University Press. [Originally published as Le latin vulgaire. Paris: Presses Universitaires de France 1967.]: 52–54). Erroneous expressions like curantes filios eius ‘while his sons were in charge’, in which the accusatives are supposed to represent an ablative absolute curantibus filiis eius, cannot be explained phonetically (Adamik 2019Adamik, Béla 2019 “Transformation of the Case System in A. L. as Evidenced in Inscriptions”. Acta Classica Universitatis Scientiarum Debreceniensis 55.13–36.: 18). The use of an ablative absolute betrays a certain amount of schooling since this construction is unlikely to occur frequently in colloquial language. Such hypercorrect attempts to follow the rules of Classical Latin demonstrate the persistence of a declensional system in the standard grammar, and at the same time a lack of proper schooling.

The written evidence suggests a scenario in which after the first language contacts between Romans and indigenous population, when more and more people were exposed to contacts with the Roman conquerors, the general proficiency in Latin improved until, eventually, the entire population became native speakers of Latin. If, however, the written sources are assigned a less conspicuous place, the contours of a different course of events become visible. If the change was not gradual, but sudden and provoked by the conquests, the early contacts between the invaders and the indigenous population must have generated reduced varieties of the conquerors’ language in the form of Latin pidgins that served as means of communication between conquerors and conquered. In due time, when the colonizers were there to stay, the next generations turned these into their native language by a process of creolization.

For the acquisition of Latin in the provinces of the Roman empire, theories of early pidginization, followed by creolization, have been advanced from time to time (Schlieben-Lange 1977Schlieben-Lange, Brigitte 1977 “L’origine des langues romanes: Un cas de créolisation”. Langues en contact, pidgins – Creoles, Languages in Contact ed. by Jürgen M. Meisel, 81–101. Tübingen: Gunter Narr.).1010.For earlier suggestions in this direction by Křepinský and Maurer see above n. 3. Gulyás (2019)Gulyás, Adrienn 2019 “How Do New Languages Arise? A comparison of romanization and gallicization”. Acta Antiqua Academiae Scientiarum Hungaricae 59.9–18. compares the adoption of Latin in the Roman empire with that of French in Martinique in colonial times, which resulted in the emergence of Kréol, a creolized variety of French. The suggestion of this scenario has met with quite some opposition. Roger Wright (2002Wright, Roger 2002 A Sociophilological Study of Late Latin. Turnhout: Brepols. : 27) rejects all explanations in terms of creolization because in his view they frame the inhabitants of the provinces as incapable of learning Latin properly. He attributes the changes in provincial Latin to a process of koineization, instead, which led to simplification and homogenization of the Italic languages when groups of soldiers brought different language varieties to the newly conquered territories.1111.In fact, creolization is one of the notions that have been proposed as replacement for that of ‘Romanization’, along with hybridization or Mediterraneanization or interconnectivity (Versluys 2014Versluys, Miguel John 2014 “Understanding Objects in Motion: An archaeological dialogue on Romanization”. Archaeological Dialogues 21:1.1–20. ). Webster (2001)Webster, Jane 2001 “Creolizing the Roman Provinces”. American Journal of Archaeology 105:2.209–225. borrows the notion of creolization from cultural studies, using it slightly differently from the way it is used in linguistics, by conceptualizing creolization as a process of blending elements from different systems. Herman (2000 [1967]Herman, József 2000 [1967] Vulgar Latin. English transl. by Roger Wright. University Park, Penn.: The Pennsylvania State University Press. [Originally published as Le latin vulgaire. Paris: Presses Universitaires de France 1967.]: 122) concedes that ‘foreigners’ learning Latin may have had a significant impact: “maybe it was they who transmitted to their descendants the language that they had learned in this unsystematic way”. He also refers to the learners’ tendency to simplify the structure of the language, linking this with the simplified structure exhibited by the Romance languages. But, he says, “this is, in the end, only a hypothesis”.

It is true that there is no direct evidence of Latin as spoken by the masses of foreigners attracted to the capital, let alone of the way it was spoken in the provinces of the Roman empire. Yet, as Robert de Dardel and Jacob Wüest (1993aDardel, Robert de & Jacob Wüest 1993a “Latin vulgaire et créolisation”. Actes du XXe Congrès international de linguistique et philologie romanes ed. by Gerold Hilty, II, 656–661. Tübingen & Basel: Francke., 1993bDardel, Robert de & Jacob Wüest 1993b “Les systèmes casuels du protoroman: Les deux cycles de simplification”. Vox Romanica 52.25–65.: 37) observe, it would be paradoxical to suppose that at first the new speakers in the provinces spoke perfect Latin and then, when the Roman Empire crumbled, progressively started to ‘corrupt’ the language. According to them, the Romance languages emerged at the very beginning of the Roman conquests when the Romans entered the provinces and communicated with the autochthonous population in a pidginized variety of Latin:

De fait, si le processus d’apprentissage du latin par des populations allophones est pour quelque chose dans la réduction des flexions latines, leur déclin a dû commencer le jour même où les premières populations non latinophones se sont mises à apprendre le latin, ce qui est absolument incompatible avec la chronologie traditionnelle, qui repose sur l’idée d’une désagrégation progressive et plutôt tardive de la déclinaison.(de Dardel and Wüest 1993bDardel, Robert de & Jacob Wüest 1993b “Les systèmes casuels du protoroman: Les deux cycles de simplification”. Vox Romanica 52.25–65.: 32)

This is similar to Křepinský’s (1958Křepinský, Maximilian 1958 “Romanica. II. La naissance des langues communes et l’existence d’une période de leur évolution commune (latin vulgaire, période romane)”. Rozpravy Československé Akademie Věd, Řada Společenských Věd 13.1–55.: 1) suggestion that the Romance languages emerged right at the beginning of the Roman conquests. De Dardel and Wüest are certainly right in drawing attention to the fact that the Romans formed a minority in the provinces and had only minimal impact on the population beyond a tiny elite. But, when they say (1993bDardel, Robert de & Jacob Wüest 1993b “Les systèmes casuels du protoroman: Les deux cycles de simplification”. Vox Romanica 52.25–65.: 37) that the emerging pidginized form of Latin could not give rise to any kind of sudden creolization since the indigenous languages remained in use for a long time after the Roman conquest, this misrepresents the nature of the process. As long as there are no native speakers, it is very well possible for a pidginized variety to be spoken for a long period of time without any tendency to creolization. Such expanded pidgins assume new functions beyond that of a crude means of communication. Eventually, often in mixed marriages, creolization sets in when a generation of native speakers grows up.1212.On the typology of pidgins and creoles see Bakker (2008)Bakker, Peter 2008 “Pidgins versus Creoles versus Pidgincreoles”. The Handbook of Pidgin and Creole Studies ed. by Silvia Kouwenberg & John Victor Singler, 130–157. Malden, Mass. & Oxford: Wiley-Blackwell. , who coined the term ‘pidgincreoles’ for those cases where the use of a pidgin variety is expanded to the point where it becomes the main language of a speech community, sometimes even acquiring a number of native speakers in the process.

As for the role of the indigenous languages, there may have been some measure of bilingualism in the auxiliary units of the Roman army in those cases where the majority of recruits came from one area, but this seems to have been the exception. In most auxiliary units, the use of other languages than Latin (and Greek in the East) was probably limited to incidental contacts between fellow soldiers who happened to come from the same region. Local bilingualism between the soldiers of the garrisons and the indigenous population must have been widespread, as evidenced by the number of loanwords in both directions. In those cases where soldiers formed more or less permanent partnerships with local women, this bilingualism may have stretched to households in the vici outside the army camp. But in the majority of cases, the ‘language of the Romans’ was bound to be the target language for children growing up in this environment because of its prestige and the tempting prospect of a profitable relationship with the Roman administration.

The net result was a population at large with varying proficiency in reduced forms of Latin, with children in mixed marriages growing up speaking a creolized variety of Latin. Only a select group of children from a small elite were granted access to Latin grammar and literature in formal education. This situation was not unlike that in other empires, such as the Arabo-Islamic empire that was established from the seventh century onwards, or for that matter, the colonial empires established by the European nations from the sixteenth century onwards.

5.Two languages instead of one

From a linguistic point of view, the concept of a general Latin speech community (Lüdtke 2016Lüdtke, Helmut 2016 Der Ursprung der romanischen Sprachen: Eine Geschichte der sprachlichen Kommunikation. 2nd ed. Kiel: Westensee-Verlag. (1st ed. 2009.): 50, 74) does not make much sense: the new speakers may have believed they were learning the language of the Romans, but for Roman native speakers most of them remained foreigners speaking a barbaric form of Latin. Within each region, a local diglossia emerged, in which those who received a formal education were able to use a standard form of Latin in writing and in formal speech, while in speaking informally everyone employed a distinctly different colloquial language. Within the empire, mutual intelligibility may have existed between some speakers from different regions, while being absent for others, depending on the level of education: an educated speaker from Iberia could communicate easily with an educated speaker from Dacia, while illiterate speakers from the two regions could not understand each other.1313.On regional differences between Latin as it was spoken in the provinces see Palmer (1954Palmer, Leonard Robert 1954 The Latin Language. London: Faber and Faber. (Repr., Norman, Okla.: University of Oklahoma Press 1988.): 174) and Adams (2007)Adams, James N. 2007 The Regional Diversification of Latin 200 BC – AD 600. Cambridge: Cambridge University Press. . On the impact of the Barbarian invasions see also Becker (2014Becker, Lidia 2014 “La protohistoire médiévale des langues romanes”. Manuel des langues romanes ed. by André Klump, Johannes Kramer & Aline Willems, 261–286. Berlin & New York: W. de Gruyter. : 272), who points out that these were not directly responsible for the fragmentation of the Romance languages, but by isolating the regions from Rome and from each other, they put an end to the unifying force of Latin and opened the gate for further fragmentation. All speakers held on to the fiction of a standard Latin, which remained oriented towards the Roman norm. The epigraphical material presents the same image of a non-differentiated Latin throughout the empire, with only marginal regional differentiation.1414. Paul Gaeng (1992)Gaeng, Paul A. 1992 “The Extent to which Inscriptional Evidence may serve as Source of ‘Vulgar,’ i.e. Spoken Latin”. Linguistica (Ljubljana) 32.19–29. seeks to reconcile the apparent uniformity of the language of the inscriptions with the notion of a locally differentiated Latin as the source of the Romance languages. He believes that statistical analysis of the inscriptions may yet bring out local differences. It should, however, be kept in mind that without taking into account the social stratification and the chronology of the inscriptions from different regions, the significance of such comparisons remains rather low. When, eight centuries after the Roman conquest of Gaul, the Council of Tours (813) recommended the use of the vernacular language (rustica romana lingua) in church, they called this language romana, apparently still regarding it as a (rustic) variety of the language of the Romans (Kramer 1998Kramer, Johannes 1998 Die Sprachbezeichnungen Latinus und Romanus im Lateinischen und Romanischen. Berlin: Erich Schmidt.).

According to Wright (2016Wright, Roger 2016 “Latin and Romance in the Medieval Period: A sociophilological approach”. The Oxford Guide to the Romance Languages ed. by Adam Ledgeway & Martin Maiden, 14–23. Oxford: Oxford University Press. : 17), the debate about the date of the emergence of regional varieties of Latin boils down to a discussion between reconstructionists and Latinists. In somewhat different terms Christine Mohrmann (1903–1988) had already remarked on the tension between those who were looking for Proto-Romance by reasoning backwards, and those who attempted to find Popular Latin in the texts (Mohrmann 1951Mohrmann, Christine 1951 “Les formes du latin dit ‘vulgaire’: Essai de chronologie et de systématisation de l’époque augustinéenne aux langues romanes”. Actes du Premier Congrès d’études classiques, 207–220. Paris: Klincksieck.: 145). Reconstructionists hold that Romance varieties were formed at an early date, whereas Latinists tend to believe that Latin survived much longer. Vincent (2016Vincent, Nigel 2016 “Continuity and Change from Latin to Romance”. Early and Late Latin: Continuity or Change? ed. by James N. Adams & Nigel Vincent, 1–13. Cambridge: Cambridge University Press. : 6–7) summarizes the debates about Vulgar Latin in the same vein. Referring to the two scenarios set up by Witold Mańczak (Mańczak 1977Mańczak, Witold 1977 Le latin classique: Langue romane commune. Wrocław: Ossolineum.), he says that it boils down to the question of whether Vulgar Latin is the sister or the daughter of Classical Latin. Just as in the debates about the questione della lingua the position one takes depends on one’s view about the demarcation between the varieties at use in the Roman empire and, after its fall, in the former provinces. Wright’s (1982)Wright, Roger 1982 Late Latin and Early Romance in Spain and Carolingian France. Liverpool: Cairns. central thesis holds that the Romance vernaculars and Latin were not conceptualized as two separate languages until the Carolingian Renaissance of the ninth century, when the imperial chancellery made a concerted effort to improve the standard of written Latin in order to create a clear difference between Latin and the local vernacular. He claims that this development began with the scripturalization of the vernacular variety, generating the need for a new orthography. Vernacular words and forms started to appear at first in Classical Latin texts, spelled more or less according to the Latin orthography (see Frank–Job and Selig 2016Frank-Job, Barbara & Maria Selig 2016 “Early Evidence and Sources”. The Oxford Guide to the Romance Languages ed. by Adam Ledgeway & Martin Maiden, 24–34. Oxford: Oxford University Press. ). But when entire texts began to be written in the vernacular, a new orthography was needed to cover all the forms. At that point, it became obvious that the two linguistic varieties were in fact distinct (Wright 2002Wright, Roger 2002 A Sociophilological Study of Late Latin. Turnhout: Brepols. : 40f.).

The awareness and conceptualization of two different languages does not preclude the possibility that the two varieties had diverged at a much earlier date. Wright is certainly right when he maintains that a linguistic change is never a matter of one structure replacing another overnight. But when he asserts that “[i]n real life all individual linguistic changes involve a period in which both the old and the new phenomenon are used, and both are understood” (Wright 2002Wright, Roger 2002 A Sociophilological Study of Late Latin. Turnhout: Brepols. : 37), the question is whether this applies across the board to all speakers: in some cases each variant may well be used by different groups of speakers (or rather: by a group of speakers and a group of learners), who do not necessarily understand both variants. In fact, they may have coexisted for a long time, just like the High and Low varieties in any other diglossic speech situation. The two varieties were not assigned to different layers of society, but all speakers shared a continuum from Low to High speech, albeit not with the same bandwidth for everyone: depending on one’s education, speakers had at their disposal a smaller or larger part of the diglossic continuum. All members of the community were convinced that they spoke the same language and pretended that the Low variety either did not exist or was nothing more than a collection of linguistic mistakes.

6.Conclusion

For a long time, the disguise of the written texts has kept the new varieties that emerged during the first contacts between Roman conquerors and indigenous population hidden from sight. One might have expected that the problematic value of written documents for the reconstruction of the spoken language would have put a stop to the notion of an intermediate language, but the idea lingers that Vulgar Latin was an actual variety of the language, retrievable from the texts and helpful for the reconstruction of Proto-Romance.1515.In this respect, the term ‘Vulgar Latin’ is used in much the same way as the term ‘Middle Arabic’ in studies of Arabic, which often serves as a label for a separate variety of Classical Arabic that is supposed to be the ancestor of the modern Arabic dialects (Janson 2012Janson, Tore 2012 “Vulgar Latin and Middle Arabic”. High vs. Low and Mixed Varieties: Status, norms and functions across time and languages ed. by Gunvor Mejdell & Lutz Edzard, 24–31. Wiesbaden: Harrassowitz.). Blau (1982)Blau, Joshua 1982 “Das frühe Neuarabisch in mittelarabischen Texten”. Grundriß der arabischen Philologie. I. Sprachwissenschaft ed. by Wolfdietrich Fischer, 83–95. Wiesbaden: Reichert. has shown that it is methodologically more correct to refer with this term to a category of texts that aim at writing Classical Arabic, but fall short of reaching this aim (see also Versteegh 2005Versteegh, Kees 2005 “Breaking the Rules without Wanting to: Hypercorrection in Middle Arabic texts”. Investigating Arabic: Current Parameters in Analysis and Learning ed. by Alaa Elgibali, 3–18. Leiden: E.J. Brill.). Significantly, the main conferences on this topic, organized since 1985 have retained the title of Latin vulgaire – Latin tardif, even though, as Väänänen (Väänänen 1981Väänänen, Veikko 1981 Introduction au latin vulgaire. 3rd ed. Paris: Klincksieck. (Repr. 2012; 1st ed. 1963.): 3) points out, one of the founders, Einar Löfstedt (1880–1955), admitted that one could never find an adequate definition for the notion of ‘Vulgar Latin’ (Löfstedt 1956 [1933]Löfstedt, Einar 1956 [1933] Syntactica: Studien und Beiträge zur historischen Syntax des Lateins, II. Lund: C.W.K. Gleerup. [Originally published, Lund: C. W. K. Gleerup 1933.]: II, 355). It is a notion that distracts from the dynamics of language acquisition in the provinces of the Roman empire. This is the fundamental insight that Cittadini and Bonamy contributed to the study of the Romance languages: what the inhabitants of the provinces were exposed to was the Latin of common soldiers and traders, often vulgar no doubt, but not the Vulgar Latin introduced by scholars to fill the gap between Latin and the Romance languages. For different reasons, each from his own perspective, Cittadini and Bonamy might have agreed with Vincent’s (2016)Vincent, Nigel 2016 “Continuity and Change from Latin to Romance”. Early and Late Latin: Continuity or Change? ed. by James N. Adams & Nigel Vincent, 1–13. Cambridge: Cambridge University Press. proposal to avoid all labels that suggest the existence of separate languages, in particular the term Vulgar Latin, and to stick to the name Latin instead. Lloyd was right when he advocated discarding the term once and for all.