How Do New Languages Arise? A Comparison of Romanization and Gallicization

This paper compares the romanization of Gaul in the 1st century BC and the gallicization of the island of Martinique during 17th-century French colonial expansion, using criteria set out by Muf- wene's Founder Principle. The Founder Principle determines key ecological factors in the formation of creole vernaculars, such as the founding populations and their proportion to the whole, language varieties spoken, and the nature and evolution of the interactions of the founding populations (also referred to as “colonization styles”). Based on the comparison, it will be claimed that new languages arise when a language undergoes vehicularization and subsequently shifts from one speech community to another. In other words, linguistic genesis would be a complicated case of language contact, where not only one, but sev- eral dialects of both superstrate and substrate varieties are involved, in a historical context where the identity function of language, or the norm, is overriden by the need to communicate. Research also indicates that language varieties spoken at the time of the shift did not pertain to normative usage, but to popular varieties, dialects, or both, since the emerging vernaculars - in Gaul, as well as in Martinique - preserved some of their phonological and lexical particularities.


INTRODUCTION
This paper is centered around the question of linguistic genesis and the epistemological issue of whether some vernaculars should be considered as exceptions to the traditional phonetic-lexical criteria used to affiliate languages genetically. 1 After a short discussion of genetic debates on the emergence of creole vernaculars and their longlasting exclusion from genetic affiliation, it will be proposed that they be considered as any other vernacular regardless of the socio-historical context they emerged in, and that the same criteria be applied to them for their genetic affiliation as to any other vernacular. A comparison of the romanization of Gaul and the gallicization of Martinique will reveal interesting parallels between linguistic phenomena accompanying the emergence of French dialects and Martinique Creole French (which will be referred to as Martiniquais later on). In sum, including creole vernaculars in the range of normalcy can be most enriching for studies on evolutionary linguistics, theories of language change, and possibly for studies on Late and/or Vulgar Latin. Evident as it might sound, voices in favor of this idea have become more prominent only in the past decade. 2

DEBATES ON LINGUISTIC GENESIS AND THE AFFILIATION OF CREOLE VERNACULARS
Ever since the 19th century there has been a trend in social sciences to draw inspiration from findings in the natural sciences. When August Schleicher developed the Stammbaum theory in 1853, he was inspired by the ideas of Charles Darwin on evolutionary biology. Just as Darwinism stipulates a common descent of all forms of life on earth, the Stammbaum theory of linguistic genesis presumes languages descend from one parent and become different from it due to geographic isolation. Although this idea originated in the 19th century -a period dominated by the "one nation, one language" ideology and White supremacy, also characterized by the stigmatization of dialects, non-standard varieties and bilingualism -the Stammbaum theory has been holding its ground ever since, influencing generations of linguists in answering the question of how new languages arise. To give a more recent example of linguists using models from the field of biology, we could mention Salikoko Mufwene's gene pool theory, 3 which uses the model of gene competition and selection to account for the competition and selection of certain linguistic features. Although the idea of monoparental genesis seems tempting in its simplicity, and biological models might reveal interesting parallels between the evolution of species and languages, linguists must not forget about the social and cultural embeddedness of language and the ensuing complexity of each and every human interaction.
Most probably as a result of the ideologies prevailing in the 19th century and continuing in the 20th century, creole vernaculars, first considered to be linguistic ---- malformations, have been denied genetic classification. The reluctance of linguists to apply the traditional phonetic-lexical correspondences criteria to these vernaculars to establish their genetic affiliation might derive from the fact that they are spoken by communities of African descent. Were these criteria applied, it would become clear that Martiniquais, for example, should be genetically affiliated to French. 4 In the trench war of creole genesis, front lines are divided as follows: Anglo-Saxon linguists are mostly substratists, and suggest Creole vernaculars descend from a proto Maritime lingua franca or from African substrate languages. They also support universalist theories such as Derek Bickerton's bioprogram hypothesis. 5 French linguists mostly argue for eurogenesis, but keep creole vernaculars out of traditional genetic classification by referring to the socio-historical conditions of their formations. As eurogenetist Chaudenson says, creolization, a "social tragedy", follows the principle of the three unities of French classical drama: the unity of action (European colonial expansion and slavery), time (about a hundred years, from the mid-17th to the mid-18th century) and place (the island). 6 The difference of opinion between Anglo-Saxon and French linguists can be justified in the sense that English-based Creole vernaculars seem to be further developed from their lexifier than their French-based counterparts, but that again might be the outcome of the different colonization styles of the British Empire and France. Also, linguists who have dealt with the question of creole genesis were and are still mostly white Europeans of the colonizing nations. Their theories singling out creole vernaculars from other vernaculars as to their genesis and affiliation might be influenced by the larger social context they came to exist in. Seventeenth-century colonization is still a relatively recent chapter in the history of humankind, involving simultaneous intermingling and segregation along racial, social and linguistic lines; moreover, it has been affecting the historical course of the colonizing nations ever since.
Understandably enough, linguists who were trying to convince the linguistic community that creole vernaculars were like any other vernacular, and therefore that they should not be defined and segregated with reference to their formation, were Black. 7 Guadeloupean romanist Hazaël-Massieux considers French-based creoles to be a second generation of Romance languages, and calls them Afro-Romance dialects. 8 Mufwene considers the socio-historical definition of creole vernaculars as discriminatory. Language contact and restructurations following contact also characterized the history of languages like English or French: why differentiate Creoles by their very definition? Congolese Mufwene proposes to define "Creole" as "a restruc-tured variety of its lexifier". 9 Along these same lines, West Indies linguist Alleyne advances that the emergence of Creole vernaculars "[…] was nothing more than the classical contact situation (Latin with Celtic, Iberian, or Italic) with differences in the degree of social integration, in the quality of the learning situation, and especially differences caused by the diverse ways in which the European/African contact situation later developed." 10 In agreement with Hazaël-Massieux's, Mufwene's and Alleyne's views, it could be stated that the group of so-called creole vernaculars cannot be delimited typologically or by any set of formal linguistic criteria. 11 In fact, when the traditional phonetic-lexical correspondences criteria are applied to Martiniquais, it is clear that Martiniquais is genetically related to its French lexifier dialects. In the following, the formation of Martiniquais will be examined and compared to the formation of French dialects after the fall of the Roman Empire.
For the comparison, the criteria set out by Mufwene's Founder Principle will be used. The Founder Principle determines key ecological factors in the formation of creole vernaculars, such as the founding populations and their proportion to the whole, the characteristics and heterogeneity of the language varieties spoken, and the nature and evolution of the interactions of the founding populations (also referred to as "colonization styles"). 12 French dialects and Martiniquais both emerged in the wake of colonization (in different periods of history, of course). From a linguistic perspective, colonization implies a multilingual or multidialectal contact situation where the language of the colonizer is first used as a vehicular, and as such, undergoes restructurations typical of vehicularization. Vehicularization is defined here as a contact situation in which the identity or group function of the language variety used as the contact language is of less importance than its communicative function. 13 A case in point is the simplification of the Latin gender system due to dialect leveling and language shift; another is the elimination of gender in Martiniquais resulting from a language shift to alloglot populations. Other examples are the preference of analytic structures to synthetic ones, the elimination of morphological or morphophonological irregularities, or the grammaticalization of word order. Despite the differences in the circumstances of colonization, it can be claimed that a language shift occurred from

COMPARISON OF THE ROMANIZATION OF GAUL AND THE GALLICIZATION OF MARTINIQUE
When Gaul was first colonized by the Romans, it was essentially a military and administrative colonization. With the Caesarean wars the Roman Empire took over Gaul, but left the organization of the colonized society mostly intact. Romans living on the territory of Gaul represented at most 500,000 people compared to about ten million natives. Romanization was slow and gradual, spanning about 500 years, spreading from Gallia Narbonensis and urban centers to more rural places and from upper classes to lower classes of society. The competence of Gaulish speakers was probably highly different, ranging from bilingual speakers in higher social classes to mostly monolingual speakers using Latin for only certain purposes. 14 The epigraphic tradition provides very little information on everyday oral interactions, but it is assumed that the linguistic reality behind the texts was extremely complex, and Gaulish society probably went through a long period of diglossia, bilingualism and plurilinguism. 15 Latin, which also became the language of the new religion, Christianity, underwent a second wave of vehicularization in Gaul with the incursions of Germanic tribes starting from as early as the 3rd century, becoming more intense after the fall of the Roman Empire. This second, Germanic colonization of Gaul involved social upheaval and ethnic crossing; still, the social prestige of the language of the Empire outlived the Empire itself, since Germanic colonizers adopted the language of the colonized. This also indicates that Latin was widely spoken in Gaul by the time of the Germanic invasions.
Another interesting finding of epigraphic studies was that certain archaic and vulgar features of the Center resurfaced in the epigraphs of the peripheries. 16 Based on these results, it can be assumed that the variety of Latin exported to the colonies and vehicularized was not literary or standard usage, but rather dialectal or vulgar varieties spoken by legionaries, many of whom were not even native speakers of Latin. These popular or vulgar varieties are conservative and innovative at the same time: they preserve features deemed archaic by standard usage, and because they are not under as much normative pressure (if any at all) as the standard, literary variety, they are also more innovative (they attest analogy in verbal paradigms, sandhi, resegmentation of lexical units, simplifications and lenition due to lax pronunciation). 1664  2722  2660  1680  3102  4900   Table 1. White and non-White population figures in Martinique in the first thirty years of French colonization by DAVID (n. 12)

Year White Non-White
As for the founding populations of the island of Martinique, colonized by the French in 1635, demographic research conducted by David (Table 1) shows that over the first thirty years of the colonization, the number of slaves never stopped growing, so that by 1680, they outnumbered White settlers. 17 Despite high mortality and low birth rates, their relative proportion grew continuously due to the influx of new slaves up until the abolition of slavery in 1848. Little is known about the exact provenance of African slaves in the West Indies, but trade gradually shifted from the West Coast of Africa (for example, Benin, Guinea), almost depopulated, towards the center of the African continent (to countries such as Congo). From 1685, contact between White speakers and slaves was restricted and marriage and cohabitation of Whites and non-Whites prohibited by the Code noir, marking the transition from the settlement period to plantation society. Despite continuing arrivals of new slaves, the use of African languages and even the use of drums (a means of telecommunication among slaves) were (at least officially) prohibited. It is almost certain that slaves illicitly used African vehiculars among themselves (maroons in the northern part of South America certainly did), but the acquisition of the vehicular used on the island, i.e., a restructured variety of non-standard, dialectal French, being inevitably necessary in order to communicate with White settlers, must have also been a practical means of communication among slaves not sharing any other common language, especially given the linguistic diversity characterizing the African continent. As far as the social and geographic background of French settlers is concerned, they were mostly poor and illiterate craftsmen and workmen from northern France, with high percentages from Normandy in particular (Table 2). Given their socio-economic background, they were most probably speakers of oïl dialects and popular French varieties. A vehicularized variety of French might have been used as an interdialectal means of communication in the West Indies among the White population. Traces of Middle French sound variations that survived in popular and dialectal varieties of French in the 17th and 18th centuries can still be detected in Martiniquais. 19 When comparing the conditions of the emergence of French varieties and Martiniquais, it is clear that even though they appeared in the wake of colonization, the colonial politics and colonization styles of the Romans and the French were very different. Pax romana represented a policy of integration, benefitting economically and culturally both the center and the provinces, leaving the indigenous social structures of the provinces intact. However, European colonization was based on political and racial segregation and was probably as disorienting an experience for White settlers as for African slaves. This might explain the fact that, whereas Romans represented at most five percent of the population of Gaul, their language was adopted by the latter. 20 There is also a considerable difference between the Roman and the French colonizations as to the duration and the intensity of contact between the founding populations. Contact between Romans and Gauls might have been less intense, but was peaceful and lasted over centuries, whereas contact between French settlers and Africans was intense at the beginning, but became much scarcer after only half a century.

Province of origin
Vehicularization was crucial both during the romanization of Gaul and the gallicization of Martinique because changes and restructurations in the arising vernaculars were initiated both by substrate varieties working as filters in the communicative process and by universal mechanisms operating in communicative emergencies: "[…] the more common or frequent, the more [sic perceptually] salient, more regular, or more transparent alternatives were favored over the less common or frequent, the less salient, the less regular, or the opaque alternatives". 21 Substrates seem to play an important role in shaping the new vernaculars by providing them with a different articulatory base and by ignoring features unknown to them: they alter intonative patterns and pronunciation, and they function as phonetic/phonemic filters. Martiniquais is, for example, insensitive to vowel length or diphthongs, compared to Canadian French that preserved long vowels and diphthongs present in 17th-century varieties of French. The series of round front vowels was also eliminated and merged with the unround front vowels. Herman also mentions how speakers of Latin in the province of Africa were insensitive to vowel length and Latin accentuation and mentions sources discussing different accents ("Spanish" or "African") for Latin speakers. 22 These phenomena demonstrate how substrate languages might change the sound shape of an arising new vernacular. Synchronic studies in spontaneous second language learning are also extremely useful in grasping the role substrate languages could have played diachronically, and how they influenced the emerging vernacular. Strategies second language learners use in a spontaneous learning context seem to result in restructurations and changes similar to the ones that can be found in languages that underwent vehicularization before their vernacularization. According to Manessy, these restructurations involve, on the phonetic level, the elimination of complex articulations (clicks, glottals, pharyngals, tones); on the phonotactic level, the simplification of consonant clusters, the elimination of hiatuses, the predominance of open syllables; on the morphosyntactic level, the elimination of grammatical redundancies and the grammaticalization of word order. 23 These changes enhance perceptibility, economy, regularity and iconicity in langage, and are all natural consequences of spontaneous second language acquisition. 24 Independently of substrate influence, vehicularization also seems to accelerate change at points in the system where there is variation already. An example of this is loss of coda Rs in Martiniquais such as palé (Fr. parler 'to speak') or liv (Fr. livre 'book'). Middle French had a tendency to eliminate both internal and word-final coda Rs. This tendency was reversed in standard French during the 17th century, but it persisted in its popular and dialectal varieties. The drop of coda Rs, probably reinforced by a tendency to open syllabification in substrate languages, terminated in Martiniquais. Martiniquais, not unlike French dialects, also retained phonetic archaisms from 16th-century French: like the variation of /o/-/u/ (koumansé Fr. commencer 'to start') or /wa/-/wɜ/ (dwèt Fr. doit/devoir 'must/have to'; mouchwè Fr. mouchoir 'handkerchief'), or the pronunciation of word-final Ts (dwèt Fr. doit 'have to', lannuit Fr. nuit 'night') and aspirated h-s (hanni Fr. hennir 'to neigh', hanch Fr. hanche 'hips') that became silent in French later on. 25 Colonial language varieties for metropolitan France, just like varieties of Latin spoken in the provinces during the Roman Empire, were, to use Nettle's metaphor, "amplifiers of variation". 26

CONCLUSION
The study and comparison of linguistic phenomena accompanying the emergence of Romance dialects and creole vernaculars reveal a great deal of similarities and led us to the assumption that linguistic genesis is not monoparental, but seems rather to be a complicated case of language contact where the language of the culturally (socially, politically) dominant community is shifted to an alloglot population. In synchrony, this situation of communicative exigency could be compared to adult second language acquisition in a spontaneous context, i.e., where the identity function of language, or the norm, is overridden by the need to communicate. These situations seem to produce very similar phenomena in different languages that all seem to increase the iconicity of language: they increase perceptibility, regularity, economy and non-marked structures. As a result of the communicative urgency that we have been calling vehicularization, a restructured version of the culturally dominant language variety emerges. The presence of archaic, dialectal and other non-standard features of the lexifier surviving in the emerging vernaculars shows that the language variety transmitted was not of normative usage, which provides valuable information as to the origin of the founding populations, their language use and the reconstitution of their contact situation.
Once social, political, and ethnic transformations settle and a new social order is achieved, the new vernacular will also mark the social identity of its speakers, develop its own norm and continue its own evolution. Once the language shift is completed and a community of speakers begins to consider the new variety as their own, vernacularization begins. Vernacularization involves the development of new pronominal or verbal paradigms, a new tense-mood-aspect (TMA) system and new syntactic procedures to express subordination or the passive voice. The process was given many a name: creolization, complexification, autonomization. The term we found the most appealing to describe the formation of a new language was Manessy's ritualization of the code -in other words, the establishment of a new norm, where the knowledge of the code and its rules has probationary functions and signals the speaker's membership in a particular socio-cultural community. 27 In contrast with many creolists who think the new vernaculars emerged in the span of forty to fifty years, this article claims, in agreement with Mufwene, that new creole vernaculars took approximately two to three hundred years to emerge, 28 just as Romance languages took centuries to form.
Acknowledging creole vernaculars as any other vernacular and classifying them as sister dialects of their lexifier language includes them in the evolutionary dynamics thereof. To classify the vernacular of Martinique as an Afro-Romance dialect includes it in the evolution of Western Romance, notably French, dialects. Thus, the emergence of new vernaculars in the wake of 17th-century European colonization and the data available on the conditions of their formation contribute to an expansion of our knowledge on how languages arise. Reconsidering old theories such as the Stammbaum theory might provide interesting, new insights on knowledge linguists already have, and might lead to different answers to fundamental questions of general linguistics. To quote paleontologist Stephen Jay Gould: "Our greatest intellectual