BACKGROUND TO ICELAND AND THE ICELANDIC SPEECH COMMUNITY

The period from c. 870 ad until the establishment of the alþingi ‘General Assembly’ in 930 ad is known as the period of the landnám (lit. ‘land-taking’). During this period, between 10,000 and 60,000 people are thought to have come to Iceland to settle. According to the historical sources the settlers came principally from southwest Norway—a part of Norway that is, even today, still dialectally rich (Íf Ii and ii, Landnámabók). We do not know what the dialectal situation in Norway was at the time, but it is unlikely that dialects did not form in Norse during the relevant period. A third of the settlers to Iceland came not directly from Norway, but from the British Isles and the Faroe Islands where they had settled previously.Footnote ⁴ There was therefore considerable so-called external colonization prior to the Settlement, that is, Iceland was settled partly by groups of people living in other Norse colonies, such as the Hebrides or the Faroe Islands, but that came originally from Norway.

The Settlement does not appear to have been an organized venture under a united leadership: the settlers took what land they could until the land became fully settled. Unlike elsewhere, notably the British Isles, the settlers did not have to accommodate to existing arrangements: they were settling in a land that was effectively empty and thus terra nullius. The settlement conditions—the settlers chose to build farmsteads around the periphery of the country and near the sea—may have led to the establishment of social structures in order that they could organize such a scattered population politically and socially. By social structures, I mean the articulation of a “set of clearly definable social institutions which were considered to constitute the basic framework of the society concerned” (Leach Reference Leach1982:32).

To speak of national languages at the time of the Settlement may be anachronistic, especially as the Scandinavians themselves appear to have been content to refer to all of them as dönsk tunga (the ‘Danish tongue’). This term refers to a “variegated picture of dialects in Scandinavia: an abstraction standing for the common elements in what were diverging dialects” (Haugen Reference Haugen1976:135). Despite the fact that the Scandinavians had the same conception of their language, and although Old Icelandic differed little from its forerunner, West Norse, there was, however, an indigenous Icelandic concern for the Icelandic language. This became explicit when writing became established. It may also be stressed that the differences between Old and modern Icelandic are generally few and insignificant, and as previously mentioned distinct dialects are virtually nonexistent.

There is a speech habit in the north of Iceland that is described as harðmæli ‘hard speech’. These speakers aspirate consonants when they occur finally or before a vowel, which is immediately preceded by another vowel or by a voiced consonant (l, m, n or ð). Alongside other small phonological variations is also the phenomenon of flámæli or ‘slack-jawed speech’, which at various times has been common in eastern and southwestern Iceland. Speakers affected by flámæli make little or no distinction between i and e or between u and ö, particularly if these vowels are to be pronounced long.

Iceland was a “focused” (Le Page & Tabouret-Keller Reference Le Page and Tabouret-Keller1985) and “closed” (Andersen Reference Andersen and Fisiak1988:60) speech community, characterized by historical norm maintenance and dense and multiplex social networks.Footnote ⁵ There is no reference to norm variation in the early grammatical treatises or in the wealth of saga literature or poetry;Footnote ⁶ if early dialect distinctions had developed, they had subsequently been lost, suppressed or were just not considered worthy of mention.Footnote ⁷

THE DIALECT SPOKEN BY THE FIRST SETTLERS WAS RELATIVELY HOMOGENOUS

Dialect formation and norm establishment in low-contact, scattered, rural speech communities such as early Iceland is problematic per se, and there are few relevant case studies.Footnote ⁸ It is even more difficult to establish the most likely scenario for dialect formation in such a distant historical period. Irrespective of when the contact took place, studies show that dialect formation tends to occur in nucleated settlements where speakers of different dialects mix. The majority of dialect leveling studies are concerned with urban, intra-territorial dialect change where the focus is on the rapid development of adolescent norms. They describe contemporary societies that have tended to discuss dialect leveling together with national and standardized languages—neither of which are appropriate in the case of early Iceland. Dialect leveling and koineization research has generally examined linguistic change characterized by high mobility, high degrees of contact, and sociolinguistic heterogeneity that involve social mixing and restructuring.Footnote ⁹

In response to the problem of how relative linguistic homogeneity came about in Iceland, Hreinn Benediktsson (Reference Benediktsson and Halldórsson1964:26) and Helgi Guðmundsson (Reference Guðmundsson1977:314–25) have argued in favor of the mixture theory, which states that the language that was taken to Iceland was a mixture of different dialects that would have been spoken in different parts of Norway and the British Isles.Footnote ¹⁰ As the name suggests, the theory claims that since the migrants came from different regions of Norway and the British Isles to Iceland, a mixture of dialects would have been spoken. Mixing—a prerequisite for leveling—requires disruption of established speech communities and the creation of a new norm. The dialect-leveling process begins with speakers “accommodating” to one another. There is the question therefore of which group of people are most likely to have adapted to others' speech in early Iceland or indeed prior to the Settlement. Studies such as Eckert Reference Eckert2000 have shown that the most innovative speakers are often adolescents, and that they are key to language change. It is difficult to find plausible contexts for the movement of large numbers of adolescents between farmsteads and isolated communities dotted around Iceland. Single individuals entering established speech communities does not lead to mixing.Footnote ¹¹

The conditions for mixing, convergence, and sustained contact were in fact largely absent in Iceland. Also, if Icelandic is a “mixed” language and the written norm was based on the spoken language, it is not clear how the norm in Iceland can be so close to the Norwegian norm. The mixture would also have included dialects of the other Norse colonies such as the Northern Isles and the Faroe Islands when some of the settlers came. This would suggest, as Kristján Árnason (Reference Árnason, Deumert and Vandenbussche2003:249) notes, that the written norm was established in Iceland and then sent back to Norway: a paradox.

There are in fact few linguistic innovations that appear in Old Icelandic before they do in the relevant West Norwegian dialects.Footnote ¹² We are thus mainly concerned with features that were either “exported” to Iceland or that developed in parallel in different parts of the Norse-speaking world.Footnote ¹³ In order to speak of dialect features in Norse at all, we must turn to the phonological structural similarities that appear in Icelandic, Faroese, and the dialects of southwest Norway subsequent to the Settlement in the thirteenth century. These include the following developments in postvocalic clusters involving sonorants: rl > dl: karl [kadl], rn > dn: korn [kodn], ll > dl: kalla [kadla], nn > dn (postlong vowel only): steinn [steidn] (cf. finna [fina]).Footnote ¹⁴

Sandøy (Reference Sandøy, Kotsinas and Helgander1994:44–45) has suggested that some of these features may be explained in structural terms. The ll > dl development only occurs in geminates, and the same cluster has been palatalized in dialects of northern Norway and certain Swedish and Danish dialects. This seems to indicate that the cluster is unstable and that it may undergo the same change in different areas without contact. It is also claimed that the rl and rn consonant clusters, which have almost the same distribution, are universally unstable and are susceptible to phonological change, undergoing assimilation or palatalization in certain Norwegian dialects. It is more difficult to see the structural motivation for the nn > dn phonological shift. Similar developments of stops before postvocalic consonants can be found in Norwegian dialects, chiefly in western Norway, in particular in the area of Gula, which lies in southwest Norway (cf. Chapman Reference Chapman1962:63–71, Lockwood Reference Lockwood1955:18–20).Footnote ¹⁵

Sandøy (Reference Sandøy, Kotsinas and Helgander1994:45, 2001:130) has claimed that these changes date back to the thirteenth century because they are not found in earlier written texts. It is reasonable to ask therefore how Icelandic can have the same structural, phonological tendencies as Faroese and the dialects of southwest Norway, if the language (dialect) spoken in Iceland was the result of dialect mixing. These changes are often attributed to “drift”, and the question of whether they can be explained by social contact or “drift” has been much discussed (cf. Björn Þórólfsson Reference Þórólfsson1925, Hreinn Benediktsson Reference Benediktsson1961–62, Chapman Reference Chapman1962, Kjartan Ottósson Reference Ottósson and Árnason2003:111–52).Footnote ¹⁶

In addition to these widely discussed consonant clusters, one might mention preaspiration. Preaspiration, the period of aspiration preceding the closure of a voiceless obstruent, is another phonological variable found in Icelandic and Faroese, but also in southwestern Norway around the city of Stavanger. The difficulty is to know what the center of origin of this particular variable might have been. Preaspiration is also found in Scottish Gaelic and so the inevitable question that arises is whether this can be attributed to a Celtic influence dating back to the Setttlement. Preaspiration is more widespread than is sometimes recognized. Pétur Helgason (Reference Helgason2002:213–28) has produced a comprehensive study of Nordic preaspiration and found it to be present in the Norwegian dialects of Jæren, Gudbrandsdalen, Trøndelag, and Senja, as well as in a number of Swedish dialects, Saami, Icelandic, and Faroese (cf. Gunnar Ólafur Hansson Reference Ólafur Hansson1997:158–64). It may be that preaspiration was geographically more widespread in Scandinavia at earlier stages than it later turned out to be.Footnote ¹⁷

Chapman (Reference Chapman1962:85) claimed that preaspiration was “an example of a feature taken to Iceland by the original settlers”. It is tempting to imagine a scenario whereby the Norsemen settle in Orkney, Shetland, the Hebrides, and so on, taking the preaspiration feature with them, and then Norwegians and Norsemen living in the British Isles (some of whom are preaspirators) settle Iceland. Irrespective of how and when this feature developed, it would seem that preaspiration is a relatively stable variable that survived leveling.

According to Chapman's (Reference Chapman1962:78) highly controversial and widely criticized monograph, these discussed sound changes are Norwegian innovations that spread to Icelandic (and Faroese) through “socio-culturally determined channels” (cf. Sandøy Reference Sandøy, Kotsinas and Helgander1994:43–45). Hreinn Benediktsson (Reference Benediktsson1963:152–62) and Ulvestad (Reference Ulvestad1964:601–4) express doubt on a number of Chapman's conclusions on chronological grounds. Phonological features such as preaspiration may be best explained by the fact that they were simply part of the phonological system of the dialect spoken in West Norway that had begun to level prior to the Settlement, and that this leveled dialect was the idiom that was taken to both the Faroe Islands and Iceland.

Icelandic does not look like a mixed language, and there are in fact sound reasons for believing that koineization might have occurred prior to the Settlement, and thus the dialect taken to Iceland was reasonably leveled. Population divergence, lack of children-peer groups and pre-established norms in early Iceland means that dialect leveling and norm establishment might be linked with significant population movements and social upheaval in mainland Scandinavia in the immediate pre-Viking period, that is, in the eighth century, a development supported by archaeological evidence (cf. Näsman & Lund Reference Näsman and Lund1988, Myhre Reference Myhre and Helle2003:83–90).Footnote ¹⁸

Considerable social change just preceding the beginnings of the Viking Age had disrupted social and geographical structures across most of Scandinavia. The roots of the koine may lie in this considerable, if undocumented, social upheaval. The koineization process may have been triggered by a radical realignment of social networks (perhaps due to social upheaval) on the west coast of Norway when the settlers came. If there were much population movement, for instance, abandonment of uncultivable land (for which there is much evidence), this would lead to population mixture, and possibly dialect leveling. The overwhelming expansion of the Scandinavian people during the Viking Age is linked with the considerable social disorder of the seventh and eighth centuries.Footnote ¹⁹ This period of internal turmoil coincides with that of a sudden and considerable linguistic change that distinguishes Norse from North Germanic (Spurkland Reference Spurkland, Fellows-Jensen, Stoklund, Nielsen and Holmberg2006:333–45).Footnote ²⁰ Social disruption of this kind is likely to be associated with linguistic change: population movement away from abandoned farm sites may permit leveling of local variation.

Looking for a place where linguistic mixing, convergence, and sustained contact may have occurred, Hedeby is an obvious starting-point, at least during its rather brief existence. It was located between Scandinavia and the Frankish empire, between Norse and Low German, near to Wendish and with direct access to the Baltic, and not that far from southeasternmost West Norse, with direct links up to the equally transient trading town of Kaupang (cf. Benedictow Reference Benedictow and Helle2003:237–49, Andersson Reference Andersson and Helle2003:312–42).Footnote ²¹ We know from Anglo-Saxon sources that English merchants also traded at Hedeby, and beyond into the Baltic (See Bately Reference Bately1980:16, the account of Wulfstan's voyage to Estonia via Hedeby (Hæðum) given in the Alfredian translation of the Orosius).

Old Icelandic appears to have developed from a unitary language, which strongly suggests a degree of leveling of any previous diversity prior to the settlers' arrival in Iceland. The reason for this is that there are likely to have been more early dialect distinctions in Norse than the runic record shows. There is an argument for koineization at the beginning of the Viking Age at the end of the eighth century. The Viking settlements in the British Isles would also have acted as dialect melting pots and thus might have been conducive to dialect leveling, prior to the Settlement. Anecdotal saga evidence indicates that the Norsemen settled in large households that were Norse speaking, even though some of their members were not Norse speakers. The Norse that they spoke did not necessarily correspond to any single location in Scandinavia. The linguistic context described here is therefore one of dialect mixing in Norway, before a process of leveling that would have continued in the British Isles, the Faroe Islands, and ultimately in Iceland itself. The Norse colonization laid the linguistic foundations for a future dialect continuum.

Dialect leveling may be seen as a gradual process; even if it occurred prior to the Settlement, this does not preclude it from continuing after the Settlement. The leveling may have begun in Norway, and then continued in the Norse colonies: the Faroe Islands were settled prior to Iceland, but its settlers, some of whom went on to Iceland subsequently, came originally from Norway and the British Isles. This created a high degree of linguistic and cultural continuity, evident in the early literature of Orkney.

Dialect formation in Icelandic appears to be similar to that of Iberian Spanish: there was and still is a dialect continuum in Spain, just as there is in Scandinavia; the first phase of southward colonial expansion in Spain occurred at a similar time to the Settlement and focused on the city of Burgos, where there was a convergence of Romance speakers from different regions across the North (Tuten Reference Tuten2003:94); koineization and leveling was a long drawn-out process; the mixing and leveling process occurred before the language was taken to the Americas. His analysis shows very clearly that koineization in the medieval period was restricted to urban or nucleated settlements. The koineization of Norse (subsequently Icelandic) is more likely to have begun in transient Viking trading towns in mainland Scandinavia like Hedeby and Kaupang than in Iceland itself.

The process of dialect formation in mainland Scandinavia appears to be quite different from the well-studied cases of dialect formation in Australia and New Zealand: in the case of New Zealand there was a different settlement pattern; the background of the settlers and social bonds between the settlers (perhaps the most significant difference between the Australia and Iceland case: convicts versus a potential elite); the nature and duration of the journey to the “new society”; much steadier and more continuous immigration in the case of New Zealand; the evidence of early social stratification in the towns; and the presence of a small Maori-speaking population meant that New Zealand was not a tabula rasa in the way that Iceland was.

There has been much discussion on the theory of new-dialect formation recently in the context of these colonial Englishes (Trudgill Reference Trudgill2004, Reference Trudgill2008:241–80), with a particular focus on the issue of the role of identity factors in language change and whether identity considerations impinge on the development of new varieties. Trudgill (Reference Trudgill2004:125) has contributed very much to this discussion, but his use of the term “identity” here appears to be restricted to “new national and colonial identities” in extra-territorial contexts—in particular Australia, New Zealand, and South Africa. It is anachronistic to speak of national identities in ninth-century Iceland unless one conceptualizes a “nation” as a self-identifying people. For this reason and because this particular linguistic development is set in the distant past and is therefore more opaque, the case of Iceland cannot contribute directly to this debate.

It would seem clear, however, that an approach that takes into account a variety of nonlinguistic factors, as Trudgill (Reference Trudgill2004:148–64) does, is beneficial to understanding the problem of relative linguistic homogeneity in Iceland. It is apparent by looking at the different dialectal situation in Iceland and the Faroe Islands that it is not sufficient just to focus on linguistic data alone.

Even within similar small-scale societies such as Iceland and the Faroe Islands, different language communities lead to different dialect convergence/divergence outcomes. Sandøy (Reference Sandøy, Kotsinas and Helgander1994:42–43, Reference Sandøy, Braunmüller and í Lon Jacobsen2001:127, Reference Sandøy, Bandle, Lennart and Widmark2005:1923–32) differentiates between two sorts of isolated, small communities: the Faroese type, where he claims that, historically, most people lived in villages of about 150 individuals; and the Icelandic type, where the population lived in isolated family units of about ten people.Footnote ²² In the Faroe Islands, a small degree of social marking of language could take place within villages, while communities remained very close-knit. This led to linguistic differentiation between, rather than within, villages, perhaps as a marker of local allegiance (see Höskuldur Þráinsson, Petersen, Jacobsen, & Hansen Reference Þráinsson, Petersen, Jacobsen and Hansen2004:339).Footnote ²³

In Iceland, there was neither social stratification nor, for the children or adolescents, any peer groups, a situation that inhibited linguistic differentiation both within a family unit and across the country itself. It is claimed that the Faroese settlement was of a different character than in Iceland, as the fewer but much larger communities represented better conditions for linguistic innovations among young people, and therefore better sociolinguistic conditions for dialectal divergence (Sandøy Reference Sandøy, Bandle, Lennart and Widmark2005:1929; see Arge Reference Arge, Mortensen and Arge2005:23).Footnote ²⁴ The dialectal heterogeneity of Faroese and the relative homogeneity of Icelandic may in part be explained by their respective settlement patterns and the type of language communities that have evolved subsequently. The study of community types is helpful when considering the scope for dialectal divergence, but it is relatively seldom that linguists refer to community types or settlement patterns as explanations of language development, such as the lack of dialectal fragmentation in Icelandic.

FACTORS THAT LED TO THE DIALECT REMAINING HOMOGENOUS

Once the variety was taken to Iceland, it appears to have remained relatively homogenous with insignificant dialect fragmentation over a long period of time. The apparent absence of local norms over a thousand-year period is problematic for three reasons: (i) as previously stated, languages tend to multiply themselves dialectally over time, a process that can take up to hundreds of years as in the case of the Englishes of Australia, New Zealand, Canada, and the US (Sapir Reference Sapir1921:160, Le Page & Tabouret-Keller Reference Le Page and Tabouret-Keller1985:50); (ii) there is clear evidence of dialectal variation in other communities that were settled by Norsemen or with whom Norsemen had contact;Footnote ²⁵ and (iii) the sociolinguistic structures that we can reconstruct for early Iceland are not conducive to one national norm, unless leveling occurred prior to the Settlement.

By sociolinguistic structures, I mean the interaction between linguistic variation (if it exists) in a certain speech community and the relevant social structures where these linguistic variants come into contact with one another. By using social network models, it is possible to reconstruct sociolinguistic structures and thus the scope for contact and accommodation. Early social structures are evidence of an incipient system of social networks, giving us some indication of the form of early social groupings in Iceland and the kind of ties that were established between individuals. The sociolinguistic structures of early Iceland indicate that contact would have been local leading to local linguistic norms, unless a norm had begun to establish itself prior to the Settlement and then was simply maintained throughout the centuries by local social structures.Footnote ²⁶

Historical sources and archaeological evidence shows that the Settlement was scattered and that the first settlers occupied large pieces of land: the peripheral settlement comprising isolated farmsteads (the Interior, the central area of the island, was (is) uninhabitable) provided few opportunities for regular convergence of all Icelanders on one place (Íf Ii and ii, Landnámabók; Ólafur Lárusson Reference Lárusson1944:16). This was combined with a lack of sustained nonlocal contact, decentralized social structures based mostly on voluntary cooperation such as local assemblies (þing) and local communes (hreppar).Footnote ²⁷ Social networks are likely to have formed on the basis of these social structures that handled problems of intra-group cooperation.

The Icelandic settlement pattern itself does not therefore necessarily support country-wide dialect leveling. Such a settlement pattern would not support a model where the language is uniform and almost dialect-free. Instead, in the absence of nucleated settlements, it is more likely to preserve dialectal distinctions.Footnote ²⁸ It would have required a very high degree of internal mobility for linguistic homogeneity to have been achieved in a short period of time. This is especially true for Iceland since there are numerous natural barriers to communication—impassable rivers, mountain ridges, glaciers and volcanoes even within the inhabited areas, barriers that one would a priori have expected to produce a much more extensive dialect splitting. Contact was based principally on local “cooperative clusters”, and thus one would have expected local norms to have established.Footnote ²⁹ Certain kinds of larger communities did exist in early Iceland, but it is nonetheless unclear how these communities would have facilitated the dialect leveling process.Footnote ³⁰

Due to the inaccessibility of the Interior, movement around the country then and now is circular.Footnote ³¹ Travelling around the edge of the country prevented the formation of isolated regions (perhaps with the exception of the West Fjords), but also rather obviously led to a circular (or perhaps semicircular) movement of people. This is likely to have contributed to the maintenance of the norm, and thus the homogeneity of the language. There is clear evidence in the sagas and the law codes of laborers travelling at set times of the year (fardagar) (Grágás Ia: 128; Íf VI Gísla saga Súrssonar: 34); movement became therefore systematized to a degree. Within the main industries of fishing and agriculture, there have always been seasonal fluctuations in employment in different parts of the country. It is also apparent from the law code, Grágás, that vagrants were itinerant, moving from one farmstead to another.Footnote ³² More significantly, there is likely to have been contact between settlers along the coast: the majority of the population lived in close proximity to the sea and some ships must have been available.

Icelandic sociolinguistic structures (in the absence of urbanization) were largely unchanged until the twentieth century and this is what differentiates it from other new societies studied by linguists. Milroy and Milroy (Reference Milroy and Milroy1985:375–84) attempt to explain this conservatism using this network theory. According to their theory, linguistic conservatism requires “strong ties” that are based on compliance with traditional norms. These strong ties are to be found in the close-knit and extended families that existed in early Iceland. Membership of certain family groups and clans was especially important in times of conflict, such as that which marked the end of the thirteenth century. In short, it was necessary to know who one's relatives were beyond one's immediate family, and it was important for other people to know with whom one was related. The difficulty with this point is that it is known that this period was characterized by phonological change, and yet the close-knit network structures that Milroy and Milroy (Reference Milroy and Milroy1985) describe would not permit that. Milroy and Milroy (Reference Milroy and Milroy1985:379) state that “Icelandic society … depended in earlier centuries on the strong networks typical of rural life. Hence, despite the difficulties of climate and terrain, social networks proved to be a cohesive force, not only in maintaining social norms, but also in maintaining the norms of language”.Footnote ³³

It is fair to assume that new societies will create new conditions for network formation as social structures become established. According to Ross (Reference Ross, Blench and Spriggs1997:223), as a community grows, there is an increase in structural heterogeneity, a gradual reduction in network density. The language of the founder community continues to be spoken in new settlements, but innovations occur in local lects. With regards to networks in Iceland, the conundrum is therefore as follows: based on Milroy's (Reference Milroy1980) use of the model and the pattern of isolated farmsteads, the probable network structure of early Iceland suggests that “local” norms would have been established and maintained. There is limited evidence of local norms, however. These networks were therefore enforcing dialectal norms that were established prior to the Settlement.

These linguistic norms have remained largely constant. High degrees of internal mobility must in part explain both the apparent absence of dialect splitting in Iceland and how the uniformity of a previously leveled dialect was maintained throughout the centuries. Internal mobility itself cannot account for norm emergence as sustained contact, and the convergence of people on one place is required. The homogeneity and conservatism of modern Icelandic is due to other factors such as Iceland's geographical isolation, the attempts to rid the language of any foreign influences, and the schooling system.

It is also due perhaps to the previous circulation of a comprehensive literature including sagas, poetry, laws, and grammatical treatises prescribing the establishment of linguistic norms. Even if the saga material is unlikely to have contributed to the dialect leveling process, it is possible that it would have been a factor in ensuring a degree of linguistic homogeneity and conservatism in the longer term.Footnote ³⁴ While they were not immutable, the sagas constituted a corpus representing the language that identified Icelanders to themselves—increasingly so as the centuries passed—and they are one of the reasons why modern Icelandic has retained its ancient language so nearly unchanged. They were written texts intended for oral performance.

It is clear already by the thirteenth century that the sagas have obtained high status and were highly valued. Kings commissioned Icelanders to write their sagas; Sverrir is the best example. Copies were made in order to be presented to kings.Footnote ³⁵ Importantly, the Icelanders read them to each other and went on reading them to each other for the next seven centuries. The Old Icelandic literature was written and preserved in centers of learning in nearly all parts of the country, without there being any noticeable difference between the texts from different areas. The language of these sagas was arguably more representative of the whole language than other registers. Instead of being representative of the whole language, the poetic, legal, and religious works would probably have been distinct registers with their separate lexicons, just as they are today.