2. Language Contact Theory.

As a starting point I wish to touch upon some problems that have long plagued contact studies. One problem is the muddle of terms and concepts used in discussing transfer processes, another is conflation or confusion of initial transfer with subsequent spread or elimination of transferred features, and a third problem is what I shall refer to as the tyranny of standard languages.

With regard to language contact theory, I employ Van Coetsem's (1988) model that distinguishes between two basic transfer types. On the one hand, RECIPIENT LANGUAGE AGENTIVITY or BORROWING refers to transfers in which the agent of transfer acts upon his own, native, linguistically dominant language. On the other hand, in SOURCE LANGUAGE AGENTIVITY or IMPOSITION the agent of transfer imposes features from his own native, linguistically dominant language on some other foreign, target language (Van Coetsem 1988:7–23). Of central importance in this formulation is the focus on the behavior of the individual bilingual as the real locus of transfer and with that the emphasis on the link between the traditional notion of linguistic stability and transfer types: Borrowing involves the least stable and least structured domains and elements of language, especially the lexicon, and imposition typically involves the more stable and more structured domains, especially the phonology (Van Coetsem 1988:25–46). Other models of transfer have failed to appreciate the importance of structural stability sufficiently and have emphasized to too great an extent social conditions, an error which ultimately obfuscates the basic distinction of transfer types.

To Van Coetsem's model Buccini (1992b) adds a third transfer type, SELECTION, which applies in situations where the bilingual is highly proficient in the two languages involved and thus linguistic dominance is in effect neutralized. The notion of selection is also useful in discussing instances of dialect contact where the structural affinity of the dialects involved is sufficiently strong that the native speaker of one dialect can easily come to be in command of another dialect. Transfer in situations of neutralized linguistic dominance does not necessarily reflect any pattern with respect to the stability gradient.

The general patterns of transfer in a given contact situation are determined according to the type of agency involved: borrowing from less stable domains in recipient language agentivity and imposition from stable domains in source language agentivity. There is, however, a separate though related process of the spread of transferred material from an original point of entry into some section of a language area, or DIALECT DIASYSTEM. The language-internal treatment of foreign linguistic features, like the original contact itself, necessarily has both important social as well as structural aspects. Failure to distinguish between the process of initial transfer and the subsequent process of dialect spread or elimination has often rendered claims about historical contacts difficult to reconcile with the nonlinguistic historical facts. Such is, without a doubt, the case in much of the work on the contact between Frankish and Gallo-Romance, and also in discussions of the English contacts with the Norse.

This pitfall of confusing original transfer with later survival and spread is, I believe, related to what I have referred to above as the tyranny of standard languages. Specifically, I mean here the common tendency to project the central status of the modern standard variety back in time, giving the standard a special existential status and creating a distorted view of the history of the dialect continuum out of which the standard arose. For example, if the dialect of Amiens or Tours had become the standard variety for France in the late medieval or early modern period, that would surely alter vulgar perceptions of the importance of the Frankish element in French but it would in no way alter the reality of what kinds of linguistic transfers were made in the course of the absorption of the Franks into the Gallo-Romance speech community in the first millennium; and the same can be said of the Norse element in English vis-à-vis the dialects of York and Winchester. If we are to say intelligent things about language contact, we must examine to as great a degree as possible the actual setting in which linguistic transfers from one language to another occurred. The subsequent issue of the fate of transferred features in broader contexts, that is, the degree to which transferred features were eliminated from or spread across the dialectal and sociolectal landscape, is obviously an important issue, but also one that can be quite distinct from and even obscure the original contact situation.

An illustration of this point is the following. One commonly speaks of the borrowing from Norse of the pronouns they, them, their into English, and from the layman's perspective, that is a sufficient formulation. From the standpoint of the specialist, however, it is an incorrect formulation that distorts the historical reality and, worse than that, leads to very wrong ideas about how transfer in language contact takes place. A far better account of these pronouns’ presence in English is that they were originally imposed by Norse speakers onto their imperfect English and eventually became the accepted forms within the Norsified xenolects of the Danelaw. As the Norsified xenolects were over time integrated into the English dialect landscape, these forms could and ultimately did spread into the dialects of neighboring areas and, through in-migration, to the more distant London, thus to areas where few or no Norse speakers had ever settled. London English, the basis of the standard, can perhaps be said to have borrowed these pronouns, not directly from Norse, but rather through dialect borrowing or selection from midlands dialects who got them ultimately as a result of imposition by Norse speakers.

The case of Norse settlement in England and the Norse linguistic influence on English, with geographically restricted imposition and subsequent spread of some Norse features outside the core zone of contact and elimination of others within that zone finds a close parallel in the case of the earlier Frankish settlement in northern Gaul. In both cases, the historical evidence for settlement patterns and the dialectal evidence together form a coherent picture: In the core areas of settlement by the conquering groups, there are obvious indications that there was sufficiently widespread imposition as to result in the establishment of partially restructured dialects or xenolects, that is, Norsified English on the one hand and Franco-Romance on the other. The two cases are also similar in that the extent of the original contact has been obscured by the fact that in both England and France the standard variety ultimately arose in areas peripheral to the zones of intense contact.

In contrast to these contact situations stands the case of French influence on English in the wake of the Norman Conquest, where there is no pattern of French influence across parts of the English dialect landscape but rather a sociolectally-determined pattern in which French lexical elements trickled down from above all across England and Scotland. And in this case, the particularly strongly affected variety became the standard.

These three oft-discussed cases of language contact can serve as the backdrop to the following discussion of a far less well-known case of language contact in medieval northwestern Europe, one in which the geographically restricted, restructured variety ultimately became a standard, namely, Dutch.

3. Absence of Secondary Umlaut and Language Contact.

Within Germanic, standard Dutch stands out conspicuously because of its extremely limited reflection of i-umlaut. In the eastern and central dialects of Dutch, i-umlaut developed essentially as it did in German; that is, there is clear evidence for the operation of both primary and secondary umlaut.Footnote ¹ In the western Dutch dialects and consequently also in the standard, there is, however, clear evidence only for the operation of the primary umlaut of Gmc. ⁺ a. Examples 1–3 show the reflexes of i-umlaut in Dutch.

1. (1) Primary Umlaut of ⁺a

   1. a.

   2. b.

2. (2) Secondary Umlaut

   1. a.

   2. b.

   3. c.

   4. d.

3. (3)

As Goossens (1980, 1988) has shown, the absence of secondary umlaut in Dutch constitutes the oldest and structurally most important feature of the language.

For the severely limited development of i-umlaut in Dutch no plausible explanation in terms of internally motivated change has been offered. Rather, Buccini (1989, 1992a, 1995, 2003) has argued that the Dutch failure of secondary umlaut is the result of language contact between Ingvæonic speakers of the coastal areas and Frankish speakers of the interior—a contact in which the Ingvæonic speakers acquired Frankish but in so doing disrupted the development of (secondary) i-umlaut. This contact occurred in the seventh and eighth centuries, when the process of i-umlaut had already run its course in Ingvæonic but had only entered the first, limited stage of primary umlaut in Frankish. In the process of acquiring Frankish, Ingvæonic speakers surely imposed various features from the stable domains of their language onto Frankish, including reduced nonprominent vowels, but likely failed to acquire non-salient, subphonemic phonetic processes of the target language. In other words, they acquired Frankish without any incipient secondary i-umlaut process, and also without i-umlaut conditioning factors in most positions. As a result, there arose a structurally independent Dutch language—a form of Frankish but one with markedly individual phonological and morphological characteristics.

The virtues of this explanation of the absence of secondary umlaut in Dutch are several. First, it obviates the need to posit undemonstrable prosodic features or claims that Low Frankish was especially peripheral in its participation in pan-Germanic structures and tendencies. Second, it links in a logical way the two most striking features of western Dutch, namely, the absence of secondary umlaut and the presence of the various Ingvæonisms. It resolves, moreover, the apparent paradox of why western Dutch, with its limited development of i-umlaut, shows so many relictal connections to Ingvæonic, precisely the branch of Germanic in which i-umlaut developed most rapidly and regularly. Third, it connects the dialectal distribution of Ingvæonisms and umlaut developments to the increase from west to east of complexity in dialectal phonological and morphological systems.

A further reason for preferring the language contact explanation is that it lays a natural basis for the complex socio-historical landscape indicated already in the prehistoric or Old Dutch period. Specifically, it is obvious that before the onset of the Middle Dutch period, dialect contact or “expansion” had taken place. Dutch, that is, the form of Frankish that was partially restructured in the course of acquisition by Ingvæonic speakers, shows from the start internal complexity and a sociolinguistic relationship between the least Ingvæonised variety, that is, the ultimate basis of the standard, and more strongly Ingvæonised varieties. Indeed, the social motivations that caused the Ingvæonic speakers to abandon their native language and acquire Frankish have later reflections that continue to this day: Once Dutch, in the narrow sense, stabilized, it has shown a continual aversion to Ingvæonic features and has been expansive at the cost of both Ingvæonic and Frankish.Footnote ²

It is clear from a cursory examination of the linguistic and historical facts that a contact of the sort sketched above could have happened. The question arises: Is there further historical evidence that would change the status of the posited contact from possible to probable? I believe there is, and in the following section I present the socio-historical evidence for the proposed Ingvæonic/Frankish contact.

4. The Dialect Landscape in the Merovingian Period.

The social and demographic upheaval that characterized the late Roman and early Medieval periods affected the Low Countries and northern France with particular intensity.Footnote ³ In the late Roman period, the population of the region decreased considerably as a result of raids by Germanic groups along the coast and from across the Rhine (see, for example, Faider-Feytmans 1964:14ff). In terms of linguistic affiliation, it seems clear that in northern France and the adjoining southern half of Belgium the Gauls and Belgae had by and large shifted from their Celtic to Vulgar Latin by 400. Following the subsequent wave of Germanic invasion and settlement of the fifth century, Germanic was ultimately absorbed into the Romance of northern France and southern Belgium and only established itself in the rest of the Low Countries, as in the German Rhineland, where the Romanized Celtic and local Germanic population was displaced or overwhelmed.

I suggest that circa 600, the core of the future Dutch language area was comprised of three major zones: First, an eastern zone of Frankish settlement, including the southeastern area around the Maas which, together with the mixed Gallo-Romance/Frankish areas in Wallonia, formed the heartland of Austrasia, soon to be the fully ascendant locus of political and military power in the Frankish world. Second, bordering the Frankish areas was an extensive, largely unpopulated zone, running from behind the coastal area of the Pas-de-Calais toward the northeast into Flanders and southern Brabant and then further northwards toward the great bend of the Rhine in the central Netherlands. This zone encompassed flooded areas in the west, the densely forested area of the Silva Carbonaria in central Belgium, and the infertile Kempen and Peel marshes in North Brabant and Dutch Limburg. The small population in the central zone was of uncertain composition but most likely by origin and orientation connected to the Franks in the east and south. Finally, to the west and north of the central no-man's-land the eastern Channel coast had been colonized predominately by North Sea Germanic speaking groups (see Gysseling 1981:104ff), both by sea as part of the general population movement that resulted in the Germanic colonization of Britain, and overland from nearby Frisian territory. This colonization was carried out primarily in the course of the fourth to sixth centuries with the Frisian expansion perhaps later. Under these conditions, which reigned roughly from the fifth to the seventh centuries, the socio-linguistic situation was quite propitious for the development of a sharp linguistic boundary between North Sea Germanic or Ingvæonic speakers on the coast and essentially Frankish speakers in the interior.

In short, much of what is now the Dutch language area lay outside the Frankish empire in the early seventh century. Indeed, until the reign of Dagobert I (623–639) the Frankish-speaking Merovingians showed little interest in the regions to the north (Petri 1937:801). When finally under Dagobert they turned northward with sustained interest in the early seventh century, missionary efforts emanating from northern France found themselves in hostile, foreign territory once they passed beyond southeastern Flanders (Lebecq 1983:48ff). Repeated missions to this area were needed before it was brought into the fold and clearly the missionary work succeeded only in conjunction with Austrasian military, economic, and political expansion (see Wallace-Hadrill's 1983:143 comments on Frankish expansion to the east). A colonial effort involving the plantation of Frankish nobility in the southern Low Countries emanating from the Austrasian Frankish heartland (Bergengruen 1958), similar to the slightly later one along the Meuse, was the crucial mechanism. In those areas across southern and northern Brabant, with a Frankish population, or at least a population speaking a form of West Germanic essentially like Frankish, the colonizing efforts of the Austrasian nobility had no great linguistic impact. But in those coastal zones where an alloglot population was being drawn into the Frankish empire and, more importantly, in the lightly populated areas between the coast and the Frankish heartland, where economic exploitation of the area by the Franks would require slave labor, the stage was set for the formation of a new, structurally distinct form of Frankish.

5. Slavery as a Setting for Language Contact.

Let us now consider the Merovingians’ motivation for conquest of and expansion into the Low Countries. First, the church desired to liberate the kindred pagan tribes from error and damnation. In addition, the expansion to the north was a means by which the king could replenish his land holdings, which were both a direct source of wealth and a crucial commodity to be given as gifts to the church and loyal aristocracy. Furthermore, the conquest of the Low Countries was also intended to check Frisian expansion and gain control of their North Sea trade.

Yet, another motivation for Merovingian expansion is of particular interest in this investigation of language contact. The conquest of territories was a way to acquire slaves who, in a time of labor shortage, were essential for the exploitation of large estates. In the wake of the Frankish conquest, estates were established in the Low Countries under royal, noble, and ecclesiastic ownership (see, for example, Ganshof 1949). On all of these domains, Frankish speakers certainly formed the leading social element. The lower rungs of the social ladder were without doubt the local Ingvæonic-speaking population, with many probably having been enslaved and others bound with onerous debts of work on those Frankish-run estates. It is almost certain too that the local labor force in this North Sea area was augmented through the purchase of English slaves at one of the nearby emporia, a practice used in the central part of the Frankish empire (see, for instance, Verlinden 1955:701). The language of the slaves from England was close enough to the local Ingvæonic dialects for the absorption of the English slaves into the larger slave population to take place. In short, in the context of slavery, speakers of the two Germanic language varieties under consideration were here brought together for prolonged periods of language contact.

Viewing the initial transfer as having occurred largely in the context of slavery also provides an explanation for the marked status of Ingvæonisms in the later history of Dutch. These features were considered low and rustic even in the later period when Flanders becomes the cultural center of the Low Countries. Finally, note that the crucial contacts and linguistic transfers need not have initially involved very large numbers of people. Indeed, it may well have been the small population of the conquering Franks that rendered the alloglot, largely enslaved Ingvæonic population proportionately large enough to have a lasting linguistic influence.

6. Early Medieval Slavery and the Church.

Since the institution of slavery has heretofore received little to no attention in discussions of language contact in the Merovingian period, it would do well to consider some aspects of slavery in general and of slavery in this early medieval context in particular. Why were the Ingvæonic speakers initially subject to enslavement, and how were they (and their speech patterns) eventually integrated into the larger speech community?

Karras (1988) provides a basic definition of a slave as a person who a) lacks all or most of the rights accorded to others in society; b) is an outsider who does not belong in a kin group or the community; c) labors under the direct control of the owner or the owner's representative; and d) belongs, in the eyes of the society in question, to a distinct status group, namely the lowest in that society. It follows logically from this definition, in which the special and alien status of the slave features so prominently, that for a variety of reasons, both practical and psychological, the institution of slavery functions better the more the slaves differ from their owners in race, ethnicity, language, and religion. Intimately connected to this principle is the fact that the most widespread and effective ways of obtaining slaves are through warfare, with the vanquished survivors being enslaved, and through trade, when one obtains the vanquished and enslaved enemies of someone else. Slavery in the Merovingian period fits this general characterization well. The vanquished Ingvæonic population, supplemented by English slaves acquired through trade, were distinct from the owners not only linguistically but also culturally since the Ingvæons were pagan.

The decline of slavery in the Merovingian context can be attributed to a combination of two factors. First, breeding of slaves is, in general, far less attractive than seizing captives by warfare, insofar as breeding requires the allocation of resources to women and children, who are in the immediate sense not productive. In the context of labor-intensive agriculture, the reduction in profits for ownership is substantial. Beyond this, slave breeding in the context of a medieval, Christian society presented a spiritual quandary. While the Catholic Church did not prohibit the ownership of slaves and was, in fact, the ultimate master of many who worked on Church-owned lands, many clergymen, as well as some secular leaders, had qualms about the ownership of Christians and, in this period of active evangelization, missionaries often turned their attention to the conversion of slaves, many of whom were then manumitted (see, for example, Bloch 1947). There was, however, a prohibition against alienation of Church property and thus a direct clash of practical and spiritual interests.

The spiritual quandary, which applied specifically to Church-owned estates, and the more general problem of maintaining a slave population given the cost of breeding and the ever-increasing shortage of pagans in the west, explain the tendency in the late Merovingian and Carolingian periods to do away with large-scale use of slave labor and instead to use serfs (see, for instance, Verhulst 1991:200ff). Serfs were responsible for the production of their own necessities but were bound to render goods and especially labor on the land exploited wholly for the benefit of the estate ownership.

One repeatedly sees at work here the principle of preferred maximal difference between owner and slave. In those instances when the difference is small, slave status is difficult or impossible to maintain. For example, English clergymen on the Continent take a particular interest in converting and freeing fellow Englishmen who are held as slaves there (see, for instance, Fletcher 1997:151ff), and more generally, we see discomfort with the idea of Christians owning fellow Christians. It seems to have been the case throughout the Low Countries that as soon as slaves were sufficiently assimilated to the culture of their surroundings, they were also soon assimilated legally and socially to the lower, peasant class of the surrounding population.

From a linguistic standpoint, this matter is important, for with the gradual absorption of the slave population into the peasant population, a sociolinguistic barrier is breached or at least reduced, so that linguistic features that had been strongly stigmatized as elements of the slaves’ xenolect can become to some degree less stigmatized elements of rustic speech and so enter into the broader regional sociolectal and dialectal diasystem.

7. Conclusion.

The language contact situation which gave rise to Dutch has been analyzed here as a case of imposition. Ingvæonic speakers in the process of acquiring Frankish imposed various features from the phonology (and other domains) of their language onto Frankish, but failed to acquire nonsalient, subphonemic phonetic processes of the target language. What began as features of an alloglot, restructured dialect subsequently spread to other dialects, including the dialect that formed the basis for the modern standard language. The origin of Dutch thus presents a parallel of the Norse origin of the English pronouns they, them, their, discussed earlier. In both, features of a language variety originally restricted in its geographic scope spread widely to become ultimately part of the standard.

In arguing for a particular scenario of linguistic transfer it is necessary not only to have a sophisticated theoretical framework elucidating language contact, but also to support the linguistic analysis with historical evidence regarding the social context of the contact among speakers. This we have provided with our discussion of the population of the Low Countries in the seventh and eighth centuries and the Frankish expansion to the north and west (see sections 4–5). Of particular interest for my purposes is the practice of slavery on the large estates established by the Franks, where Ingvæonic-speaking slaves and serfs found it necessary to learn the language variety spoken by the more powerful Franks. The intermediate position of Dutch between English and German can then be understood in large part as the result of the intimate sociolinguistic interaction of Ingvæonic and Frankish in the late Merovingian and early Carolingian periods.