Multilingual practices in late medieval Swedish writing
Steffen Höder (Kiel University)
Late medieval Sweden was a multilingual society. At least three languages – namely Old Swedish, Low
German, and Latin – were in use, beside other regional languages. While the influence of Low German is
easily detectable in all parts of the Swedish language system and has been investigated rather thoroughly
from a historical sociolinguistic point of view (cf. Braunmüller 2004), the role of Latin has been rather
marginalized in traditional Swedish language historiography, focusing on the earlier stages of Old
Swedish, which are described as its classical form (cf. Pettersson 2005). Starting out as the language of
religion, administration, diplomacy and, to some extent, trade, Latin was the dominant language of text
production in Sweden until the 14th century, which saw Written Old Swedish gain some domains as well,
resulting in a more balanced diglossic relation between the two languages. The emerging written variety
of Swedish, however, was heavily influenced by the multilingual practices of scribes, in large part clerics
who were used to using at least Swedish and Latin on a daily basis for a variety of communicative
purposes (Höder 2010). These multilingual practices, ranging from ad hoc translations via code-switching
to the application of Latin stylistic, textual, and syntactic norms in Swedish text production (Höder 2018),
had a lasting impact on the later development of a Swedish proto-standard, and are still reflected in
conservative text types today. This contribution approaches this development from a historical
sociolinguistic and contact linguistic perspective, concentrating on the establishment of multilingual
Keywords: Old Swedish; Latin; standardization; language Ausbau; multilingual practices
1 Old Swedish in traditional language historiography
The history of a language is not the same as the history of its speaker community or communities. This
distinction is different from the more traditional dichotomy between a language’s internal and external
history, i.e. the history of linguistic structures as opposed to social, demographic, and other historical
parameters that have influenced the structural development. Rather, it points to the insight that a
‘linguocentric’ approach to language history tends to miss some central points of a language’s ecology, to
use the term introduced in Haugen’s (1971) seminal paper (for a comprehensive overview, cf. Eliasson
2015), such as: Who used language X at a given point in time? In which domains? Was X a spoken and/or
a written language? Was it standardized? What were the attitudes of its speakers towards X? And,
crucially: Was X the only language used by its speakers, or were they bi-/multilingual?
Language historiography, however, remains largely conservative and linguocentric in its take on
language history. One example is the historiographical approach to Old Swedish (OSw; Swedish
fornsvenska), i.e. late medieval Swedish, conventionally dated to the period between c. 1225, when the
oldest extant Swedish text written in Latin script was produced (Äldre Västgötalagen, the Elder
Westrogothic law), and 1526, when the first Swedish translation of the New Testament was printed. OSw
is usually described in histories of Swedish, not histories of Sweden or of the Swedish-speaking
people(s). This implies not only a focus on language-internal developments (internal history), but also a
rather pervasive monolingual perspective on structural developments and their consequences – while, as a
matter of fact, late medieval Sweden was to a large extent a multilingual society (cf. Section 2).
The monolingual perspective is evident in, among other things, the traditional emphasis on Early OSw
(roughly, the first half of the whole OSw period, 1225–1375) that is characteristic of older standard works
on language history and Old Swedish grammaticography (e.g. Noreen 1904, Wessén 1941–1956). This
focus on Early OSw may in part be traced back to the roots of language historiography in 19th-century
philology, which – quite naturally in the age of national romanticism – considered the oldest attested or
reconstructible linguistic structures to represent the most genuine form of a language and, thus, the one
most worthy of analysis. (Early OSw preserves far more of the inherited Nordic or Germanic inflectional
morphology than does Late OSw, which shows strong deflectional tendencies as well as other types of
structural change attributable to ‘foreign’, i.e. primarily German, influence.) The traditional emphasis on
Early OSw is still reflected in more recent works such as Pettersson’s (2005) textbook on Swedish
language history, when the earlier period is labelled klassisk fornsvenska ‘classical OSw’ – an evaluative
and idealising rather than a descriptive term. In effect, this concentration on the oldest and most genuine
forms almost inevitably entails a marginalization of the role of other languages in medieval Sweden, and
an invisibilization of these other languages in the history of Swedish (sensu Langer & Havinga 2015).
Moreover, it is very much debatable whether ‘Swedish’ at the beginning of the 13th century actually
should be thought of as a language in the modern sense that the terminology seems to imply. In many
respects, that would be quite an anachronistic view: Rather, the historiographical label ‘(Early Old)
Swedish’ refers to a geographical bundle of East Nordic dialects within the Scandinavian dialect
continuum, increasingly accompanied (but not yet roofed) by a set of emerging writing traditions, not
clearly delimitable from the neighbouring Nordic languages (Danish, Norwegian, Gutnish) on structural
grounds, defined only in political terms as the Nordic language as used within the recently consolidated
Swedish realm. Even the use of the term svenska ‘Swedish’ from the 14th century onward cannot be taken
as clear evidence for the existence of any kind of linguistic awareness or focusing (sensu Le Page &
Tabouret-Keller 1985) beyond the fact that the vernacular Nordic dialects were perceived as different
from allochthonous languages (such as German; cf. Ottosson 2002).
Of course, an integrative approach, following in Haugen’s footsteps in many respects, has come to
play a major role in historical sociolinguistics and contact linguistics. For instance, a multilingual
perspective figures prominently in historical sociolinguistic work on language contact between medieval
Swedish (as well as other Nordic languages) and Low German, and in particular the work by Braunmüller
and associates has indeed become quite influential (for an overview, cf. Braunmüller 2002, 2004, 2005).
While this line of work deserves credit for addressing the role of language contact in language change,
and for putting historical multilingualism on the map in the first place, it still occupies a rather specialist
niche, and even if contacts between Swedish and other languages are accepted as an important part of
language history (cf. Teleman 2002: 24–41), a multilingual perspective is still far from belonging to the
historiographical mainstream. Furthermore, our picture of how important multilingualism was for
medieval Sweden as a linguistic ecosystem is yet far from complete.
This article embraces the view put forward by, among others, Pahta, Skaffari & Wright (2018: 4) that
“one way or the other, virtually all historical texts are multilingual” in the sense that they are – in many
cases rather heavily – characterized by what can be labelled ‘multilingual practices’, and that this textual
multilingualism reflects the fact that historical societies usually were multilingual, too. The article
contributes some findings on multilingual practices employed by speakers of Latin and Swedish and
argues for their place in Swedish language history – both in the sense of a history of Swedish and in the
sense of Sweden’s linguistic history. Section 2 discusses the Ausbau of OSw, i.e. its development into a
written language, in the multilingual society of medieval Sweden. Section 3 focuses on two case studies
of Ausbau-related structural innovations that exemplify different fates of emerging multilingual practices
in the later development of Swedish from the Late Middle Ages onwards. Based on these findings, the
conclusion in Section 4 reassesses the role of multilingual practices in Swedish language history.
2 Ausbau in a multilingual society
While traditional language historiography tends to focus on Early OSw for reasons outlined above, it is
clearly Late OSw that has had most influence in the later development of Swedish, in particular with
respect to its use as a written language. Firstly, Present-Day Standard Swedish is the outcome of a
standardization process starting in the Early Modern Swedish period (after 1526) and culminating in the
work of the newly founded academies in the 18th century as well as the codification of a written standard
in 1801 (Teleman 2002: 69–71; 2003: 406–407). However, written Early Modern Swedish also built on
earlier writing practices established in Late OSw, particularly the traditions of writing in the then
dominating genres of religious prose, which can be viewed as constituting a proto-standard in Deumert &
Vandenbussche’s (2003: 456) terms (Teleman 2003: 409; cf. Figure 1).
Figure 1: The place of Late Old Swedish in the history of Standard Swedish
Secondly, Late OSw is by far the most dominant period in the textual history of OSw, both in quantitative
terms, with a sharp rise in text production during the 15th century, and in qualitative terms, with a
significant diversification of the types of texts produced in Swedish, accompanied by a gradual
development towards a more unified written norm. This boost in text production is mainly the result of an
ongoing process of language Ausbau, to use the term coined by Kloss (1967, 1978). Ausbau can be
defined as a historical process leading to a relevant group of speakers of language X writing texts in
socially important domains in X, as opposed to earlier practices of writing in other languages or not
writing at all (Höder 2010: 76–78). Ausbau normally – and certainly in European history – presupposes
language contact and at least some degree of multilingualism within the relevant speaker group: People
will seldom develop writing from scratch, but rather develop writing practices in a formerly unwritten
language based on their previous experience with writing in other languages, and there will hardly be an
abrupt switch from writing in language Y to writing in X, but rather a gradual shift, either towards writing
exclusively in X or towards writing in both languages, at least as an intermediate stage. Also, Ausbau
tends to proceed domain-wise, with the Ausbau language gaining some domains earlier than others.
The Ausbau of OSw started off around 1300 in a polyglossic situation (Figure 2). Sweden as a whole
(then including today’s Finland) was a multilingual society, where (a) Swedish was used alongside a
1200 1300 1400 1500 1600
Early OSw Late OSw
range of other languages, including on the one hand (b) Uralic languages spoken by regional minorities in
some regions (Finns, Sami), and on the other hand (c) Low German and (d) Latin. While regional
minority languages were usually not written until much later (but see Blomqvist 2017 on traces of Finnish
in OSw documents), Latin and Low German did not only function as highly developed written languages
in medieval Sweden, but, more importantly, were also highly prestigious, mainly as a consequence of
their association with prestigious domains in Swedish society. While Swedish was mostly used in
everyday domains by the majority of the population, and primarily as a spoken language (with the notable
exception of some early text types, such as the so-called provincial laws), Low German was chiefly used
in the domains of trade and town administration, in its function as the lingua franca of the Hanseatic
League, and in everyday domains by the numerous German minority living in Swedish towns.
contrast, was used in the domains of the church – including religious domains in a narrower sense, but
also e.g. ecclesiastical and monastic administration –, in administrative domains throughout the country,
in religious and secular literature, and in communication with foreigners as the lingua franca of Western
Europe (Höder 2010: 28–38).
Figure 2: Swedish polyglossia (around 1300)
One of the first triggers for language Ausbau was what can be described as a first modest step towards
an official language policy: a political attempt to make Sweden’s administration more efficient by
regulating the language used in public charters. The first law for the entire Swedish realm (Magnus
Eriksson’s Country Law, Magnus Erikssons landslag, passed around 1341 and superseding the earlier
provincial laws), stipulated at least for the countryside that all public charters must be written in
A similar, albeit not exactly parallel, provision in the contemporaneous town law (Magnus
Eriksson’s Town Law, Magnus Erikssons stadslag) stated that town clerks should be Swedish and that
Swedish members should constitute at least half of the town councils (Larsson 2003: 57). One effect of
these regulations was to strengthen the position of Swedish as a legal language in towns as well. Indeed,
the proportion of Swedish-language charters increased from 5 per cent in the decade before 1350 to 30
everyday domains (Germans)
lingua franca (Hanseatic League)
lingua franca (Europe)
primarily spoken, emerging written variety
per cent one decade and 71 per cent seventy years later (Gejrot 2011: 101). In Ausbau terms, this means
that Swedish gained a domain from Latin, and bilingual writers wrote increasingly in Swedish instead of
(or in addition to) Latin.
Apart from charters, the most important trigger for language Ausbau was the text production of
monastic communities (Höder 2010: 85–87), above all the Birgittine order with its motherhouse Vadstena
Abbey, situated on Lake Vättern in southern Sweden, and several other houses throughout the country.
The Birgittines, initially founded by Birgitta Birgersdotter (Saint Bridget) in 1344, were an unusual
religious order in several respects: Firstly, the order had both male and female members; Vadstena Abbey,
for instance, was a double monastery with a couple of monks and a larger number of nuns. Secondly, the
order – and Vadstena Abbey in particular – soon assumed the role of Sweden’s leading spiritual and
cultural institution, and even exerted political influence. Thirdly and most notably, the Birgittine
monasteries became the centre of monastic text production in Latin as well as in the vernacular Swedish.
The high proportion of religious prose texts, the predominant genre during the Late OSw period, reflects
the level of the Birgittines’ productivity. This is illustrated in Figure 3 (based on Wollin 1991a: 246, Fig.
1), which shows the proportion of different genres in extant OSw sources (excluding charters); whereas
the high number of religious texts is due to the production of original texts and translations, the
comparatively high number of laws is based on the existence of many copies per text (cf. Åström 1993:
Figure 3: Old Swedish text production across text types
Belonging to a small, societally influential, highly educated cultural elite group, the Birgittines also
formed part of a tight-knit community of practice (sensu Meyerhoff 2002, cf. also Timofeeva 2013 on a
similar case with Old English-Latin bilinguals), in which different activities connected to literacy were
part of people’s everyday life: The Birgittine monks and nuns were by no means novices to writing, but
rather experienced, if not professional clerks, scribes, authors, and not least translators (and described as
such in, for instance, Vadstena Abbey’s memorial book; cf. Gejrot 1988). Furthermore, most members of
the Birgittine community were at least bilingual to some extent: Being an experienced writer was, in
principle, equivalent to being an experienced Latin writer, as Latin still was the primary means of written
communication, and Latin and literacy were, even in formal training, inextricably linked to each other. As
a consequence, Swedish writers could rely on a tradition of firmly established textual, stylistic, and even
syntactic norms when writing in Latin, whereas a corresponding set of specifically Swedish norms had
not yet been developed. It is not hard to imagine that bilingual writers, in the absence of specifically
Swedish norms, did not necessarily view the Latin ones as applicable to one language only, and rather
interpreted them as norms for writing as such, in whatever language (cf. Höder 2018: 158–159). What is
more, there is even metalinguistic evidence that writers were actively encouraged by their superiors to
employ Latin norms when writing in Swedish. For example, a bishop wrote to the community at Vadstena
Abbey in c. 1495 that a good preacher was supposed to “improve this dark Swedish according to the Latin
3 Corpus analysis: Multilingual practices in HaCOSSA-late
We see in the Swedish Birgittine texts an abundance of various forms of multilingual practices, ranging
from classic code-switching phenomena (such as insertional and alternational code-switching between
Swedish and Latin) to different structural innovations that can be explained as representing contact-
induced change (see below). For some of these innovations, more specifically, it seems plausible that they
originated in or at least facilitated the communicative task of translating from Latin into Swedish (on the
role of translations in contact-induced language change, cf. Kranich, Becher & Höder 2011). Many of the
Birgittine texts were translations, increasingly aiming at formal equivalence between the Latin originals
and their Swedish counterparts (instead of the earlier practice of achieving functional equivalence by
paraphrasing where necessary; cf. Wollin 1991b). The more closely OSw structures resembled the Latin
model, the easier it was to produce formally equivalent translations. In addition, the Birgittine tradition
included the monks’ preaching to laypeople in the vernacular, based on sermons drafted in Latin (Wollin
2001: 142–143) – the monks’ sheer ability to do that points to an impressive degree of routinization of
some multilingual practices within the monastic community.
The following sections present findings from a corpus analysis of Late OSw data. The corpus is
presented as a multilingual corpus in Section 3.1. Section 3.2 discusses innovative features of relative
clauses in Late OSw and argues for an explanation based on Latin-Swedish bilingual practices in
monastic communities. While these innovations caught on in the later development of Swedish and made
their way into the written variety of Modern Swedish, others did not. Section 3.3 discusses one of those
multilingual dead ends, namely the short-lived OSw gerundive.
3.1 A multilingual corpus
The corpus used is The Hamburg Corpus of Old Swedish with Syntactic Annotation (HaCOSSA), an
XML-based digital corpus of OSw texts, manually annotated syntactically and morphologically according
to TEI, Menota, and PaCMan standards.
The corpus comprises different text types and contains about
113,000 tokens in total. The analyses in this article are based on HaCOSSA-late, a sub-corpus containing
eleven Late OSw texts with 80,586 tokens in total (cf. Table 1 for details). While not all of the texts have
a religious content in a narrow sense, although religious fiction and sermons make up most of the corpus,
the remainder also belongs to the monastic sphere, either because the texts were used in the Birgittine
administration or, in the case of the two secular fictional texts, because they were written or translated by
members of the monastic community.
Heliga Birgittas uppenbarelser,
book 4, chapters 1–20, 40–60
Heliga Birgittas uppenbarelser,
Heliga Mechtilds uppenbarelser,
introduction and Second Commandment
Sermones sacri Svecice
Ordning vid val af
Confessor Generalis i Vadstena kloster
Stadga af år 1443
för Vadstena klosters ekonomi
Handlingar på svenska rörande ’Vårfrupänningen’
till Vadstena klosters byggnad och underhåll,
Aff Joan prest aff India land
Table 1: HaCOSSA-late
A convenient way to establish that this corpus represents multilingual practices that were typical of
monastic text production during the Late OSw period is to make sure it contains (overtly) Latin material,
i.e. code-switching phenomena. It is crucial, though, that the Latin elements do not (solely) consist of
established loanwords, but also exhibit Latin grammatical marking, and thus indicate the productive use
of Latin alongside Swedish by the writers.
While code-switching as such is not annotated in HaCOSSA and, therefore, hard to quantify, the
number of Latin tokens (which are annotated) may serve as an approximation in quantitative terms.
Table 2 shows the normalized frequency of Latin tokens in HaCOSSA-late per text type:
(per 1,000 tokens)
Table 2: Latin elements in HaCOSSA-late
While the frequency of Latin elements varies greatly across text types, ranging from low figures in
religious and secular fiction (12.4 and 16.0 per 1000 tokens, respectively) to an exceedingly high figure
(225.0 per 1000 tokens) in sermons, it can be established that, on average, approximately one in thirty
tokens is Latin.
As to the forms and functions of code-switching in Late OSw, there does not seem to be much of a
difference between the strategies found here and in other data, both historical and more recent; in
particular, code-switching is employed in connection with lexical gaps and cultural loans, but also as a
means of marking textual and discourse units as well as quotations, in part irrespective of the original
language (cf. Höder 2018: 155–157). While it is obvious that the production of written texts allows for a
higher degree of planning and even correction, and while it is plausible that this might suppress some
forms of unintended code-switching that occur more frequently in oral speech, code-switching in
multilingual texts does not generally appear to be fundamentally different from code-switching in spoken
language (Gardner-Chloros 2018: 21–24), and more recent work as well as classic studies such as Stolt’s
(1964) analysis of Luther’s Table Talks and Wright’s (2002) study on medieval business texts have
yielded similar results.
The most notable formal feature of code-switching in HaCOSSA-late is the use of Latin inflectional
morphology in insertional code-switching. For example, (1) – taken from a text on procedural rules for
the election of the general confessor at Vadstena Abbey – contains several Latin lexemes (marked by
(1) CG 113:
Ok nar han som waldir är til confessorem a sinne matto som
and when he as elected is to confessor-ACC.SG on his manner REL
fore saght är tha halde han confessoris stadh ok ämbite
before said is then hold-SBJV he confessor-GEN.SG place and office
thogh ey som confessor wtan som en slättir brodhir
though not as confessor-NOM.SG but as a simple brother
thogh skula alle hanom lydha som i Regula saluatoris
though shall all him obey as in rule-ABL.SG saviour-GEN.SG
biudz lydha confessori i xii capitulo
command-PASS obey confessor-DAT.SG in 12 chapter-ABL.SG
“And when he has been elected as confessor in the aforementioned manner, he shall take the
confessor’s place and office, yet not as the confessor, but as a simple brother. Yet all shall obey
him as it is commanded in the twelfth chapter of the Saviour’s Rule to obey the confessor.”
In (1) there are four instances of Latin confessor ‘confessor’, each in a different inflectional form. This
lexeme is a technical term, meaning – in this context – the general confessor of Vadstena Abbey. The
different case forms, marked by the appropriate Latin suffixes, are governed by Swedish elements
according to Swedish morphosyntactic patterns: The accusative confessorem is governed by the
the genitive confessoris specifies the possessor in a possessive noun phrase, the
nominative confessor preceded by the particle som ‘as’ agrees with the subject of the clause (the 3rd
person singular pronoun han), and the dative confessori is governed by the verb lydha ‘obey’. Similarly,
the ablative form capitulo is governed by the Swedish preposition i (for the rather intricate matter of
conventional equivalence between OSw and Latin cases, cf. Höder 2012: 250–253). Whether Regula
saluatoris, the ‘Saviour’s Rule’ can be counted as code-switching or as an established loan, is impossible
to decide; as a fixed expression, it denotes the basic rule of the Birgittine order.
Similar examples abound in HaCOSSA-late, suggesting not only that the writers had an intimate
knowledge of both Latin and Swedish and were productively bilingual, but also that code-switching was,
at least to a certain degree, an established multilingual practice within the monastic community.
3.2 Monolingually surviving innovations: new features of relative clauses
Apart from code-switching, reflexes of multilingual practices can be seen in structural innovations such as
the establishment of new features of relative clauses. This particular innovation, singled out for discussion
in this article, can also be seen as a part of a larger pattern of structural innovations that all have to do
with a tendency towards a more explicit way of clause-linking and, in many cases, a gradual development
towards a binary distinction between co- and subordinating structures in Late OSw. This involves among
other things the emergence of semantically unambiguous subjunctions, specifically subordinating word-
order patterns, and absolute participial constructions. All of these innovations are linked to the ongoing
Ausbau of OSw (for a comprehensive study, see Höder 2010).
In general, Old Nordic adnominal relative clauses (RCs) follow their antecedent (not considering free
RCs without an antecedent). Old Nordic languages, including OSw, normally mark RCs in one of two
ways: (a) by a relative subjunction, i.e. an uninflected clause-initial particle (such as OSw sum or þär; cf.
Modern English that) that does not carry any additional grammatical meaning, or (b) by zero, i.e. the
initial element in the RC is either its subject or – when the antecedent functions as subject within the RC
– another constituent. These two strategies are illustrated in (2) and (3):
(2) BA 74:
þin vikarius þær sittar i þinum staþ
your deputyLat REL sits in your place
“your deputy that sits in your place”
(3) JP 345:
och thw wilth wetha the vndher Ø i vaara lande ärw
and you want know the wonders (REL) in our country are
“and you want to know the wonders that are in our country”
In Late OSw texts, RCs can also be marked (c) by pronominal elements, i.e. forms of relative pronouns
(such as hviliken or þän) inflected for gender and number (in agreement with the antecedent) as well as
case (governed by some element within the RC; cf. Modern English who vs. whom), or (d) by a
morphologically complex relativizer consisting of a relative pronoun and a relative subjunction (e.g.
hviliken sum). These additional strategies are illustrated in (4) and (5):
(4) BL 7.8 (162):
thet klædith om hwlkith hans korsfæstare dobbladho
that robe about REL-N.SG.ACC his crucifiers threw.dice
“the robe for which his crucifiers threw dice”
(5) BL 7.7 (160)
mangha pawa waro før iohannez pawa hwlke som føro till hælwitis
many popes were before John pope REL-PL.NOM REL went to hell
“there were many popes before Pope John who went to hell”
While the older, non-pronominal strategies still make up more than half of all adnominal RCs in
HaCOSSA-late (962 of 1,598 instances, or 60.2 per cent), the innovative RC markers that involve a
pronominal element appear to be firmly established in written Late OSw as well (636 instances, or 39.8
per cent). Pronominal relativization was unknown in earlier Old Nordic, appears only marginally in Old
Norse (Faarlund 2004: 264–265), and is typologically rare (Comrie 1998, Comrie & Kuteva 2013). It is,
however, the only RC marking strategy of Latin (the default relative pronoun being qui). As mentioned
above, the existence of an OSw RC marking strategy that is structurally isomorphic to the strategy
employed in Latin must have greatly facilitated the task of translating Latin texts when aiming at formal
What is more, language contact or, to be more precise, multilingual practices as used by Latin-Swedish
bilinguals appear to have been a decisive factor in the very emergence of pronominal relativization in
OSw. The emergence of relative hviliken has indeed often been attributed to Latin influence (e.g. Wessén
1941: 78–79 [§113–114]). Höder (2010: 217–219) interprets it more precisely as the result of, in Heine &
Kuteva’s (2005) terms, a process of ‘replica grammaticalization’. Replica grammaticalization is a process
in which multilingual speakers notice the existence of a particular category X in one of their languages
and then create a corresponding category in another language by replicating a grammaticalization process
that has resulted in the emergence of X. In this case, it is assumed that Latin-Swedish bilinguals identified
the Latin relativizer qui with the homophonous interrogative determiner qui and its Swedish counterpart
hviliken and subsequently grammaticalized Swedish interrogative hviliken into a relative pronoun.
Whatever the explanative value of such an approach, it is obvious that the use of hviliken as a relativizer
entails an increase in structural similarity between the two systems (cf. Höder’s [2012: 253–255]
Diasystematic Construction Grammar approach to this change, which advocates the view that the Late
OSw relativizer inventory is best described in terms of a common system shared with Latin).
The emergence of the relative pronoun þän is a different, but related story. At first glance, it is the
result of a largely language-internal grammaticalization process, which involves a reanalysis of sequences
consisting of demonstrative antecedents with subjunctionally or zero-marked RCs as pronoun-initial RCs.
While there are unambiguous cases where forms of þän must be analysed as relative pronouns, there are
also, as one should expect, abundant cases of ambiguous contexts in which either analysis can apply, as is
illustrated in (6):
(6) BL 7.8 (162)
alla the som sægya pawan ey vara sannan pawa
all ?-PL.NOM REL say pope-DEF not be true pope
“all those who say that the pope is not the real pope”
The sequence the som can either be composed of a demonstrative pronoun (the, plural of þän) followed
by a relative subjunction (som) – and, hence, as subjunctional RC marking –, or it can be a
morphologically complex relativizer (relative pronoun plus relative subjunction; cf. Höder 2010: 212–
213). Even as a language-internal process, however, the grammaticalization of þän as a relative pronoun
results ultimately in an increasing structural similarity between Latin and OSw. It is, therefore, not
unlikely that it has at least been reinforced through multilingual practices, and particularly the practice of
formally equivalent translation.
In addition to the innovative formal RC marking strategies, there are also innovations related to the
function of RCs in OSw texts. Generally speaking, adnominal RCs can be either restrictive or non-
restrictive. Restrictive RCs restrict the reference of their antecedent (e.g. I really like the town that I live
in), while non-restrictive RCs just contain additional information about it (e.g. I really like the town which
has a population of 300,000). While restrictive RCs are generally more frequent than non-restrictive ones
in HaCOSSA-late (890 of 1,598 instances, or 55.7 per cent), non-restrictive relative clauses are still
remarkably frequent (708 instances, or 44.3 per cent) as compared to numbers for earlier texts (Höder
2010: 203–205). Whether or not non-restrictive RCs as such are a structural innovation in Late OSw is
hard to determine, the chief reason being that it is not always possible to unambiguously classify
individual RCs, since the classification has to rely on a more or less interpretative reading of a clause in
its context. Obviously, however, non-restrictive RCs represent a fairly frequent clause-linking strategy in
Latin, and were usually translated into formally equivalent RCs in Late OSw (Höder 2010: 216–217).
Interestingly, the two innovative features that are likely to be associated with the establishment of
multilingual practices during the Late OSw period – pronominal RC marking and non-restrictive RCs –
are also related to each other. Table 3 shows the distribution of pronominal and non-pronominal markers
across restrictive and non-restrictive RCs in HaCOSSA-late:
pronoun + subjunction
Table 3: Innovative RC types and marking strategies in HaCOSSA-late
The pronominal markers occur more frequently with non-restrictive RCs than with restrictive ones,
whereas restrictive RCs tend to be marked non-pronominally (p < 0.01, χ² test) – in other words, writers
tended to mark RCs in a manner that was structurally similar to Latin, and even more so with RCs that are
also functionally similar to a prominent Latin type.
Innovative RC features in Late OSw represent, in a manner of speaking, a successful change: Both
relative pronouns and non-restrictive relative clauses caught on in the later development of written
Swedish. Starting off as (a result of) multilingual practices at a stage of Swedish language history when
text production was dominated by multilingual writers, they have survived into the later monolingual
history of Modern Swedish (at least in written registers), even though writers in later periods were not
necessarily bilingual any longer, or at least not to the same extent as the monastic writers of the Late
Middle Ages. The development of written Swedish in the 20th (and 21st) century has, however, been
characterized by a tendency to minimize the distance between written and spoken language. Whereas
relative pronouns (unlike complex relativizers) still exist in Present-Day Swedish and can be found in
grammatical descriptions, they are today usually restricted to rather formal, conservative text types in, for
instance legal and religious domains (cf. SAG, 4, 492–494 [§22], Pettersson 1976).
3.3 Multilingual dead ends: the gerundive
While there are successful innovations that can be traced back to multilingual practices during the Late
OSw period, not all of the innovations that came about in a similar way were as successful – there are also
what can be called multilingual dead ends. A telling example is the Late OSw gerundive.
Originally, Old Nordic, including OSw, did not have a specific inflectional form of transitive verbs to
express a passive obligation (i.e. the idea that something should be done to a patient), although,
occasionally, an active participle would be used to express passive obligation (Wessén 1956: 152 [§ 152]).
Latin, on the other hand, has a verbal adjective called the gerundive which is regularly formed by adding
a suffix -nd to the verb stem and which in itself is inflected as an adjective (in agreement with a noun
denoting the patient of the action referred to), such as delendus ‘[someone/something] that has to be
destroyed’ (< deleo ‘I destroy’).
This structural mismatch between Latin and OSw posed something of a problem for monastic
translators aiming at formal equivalence between Latin original texts and OSw translations. Late OSw
corpus data indicate that translators solved this problem in different ways. One alternative was, obviously,
to give up formal equivalence and find a different translation for a Latin gerundive. More interestingly,
however, OSw translators actually also attempted to use morphologically corresponding forms in order to
achieve formal equivalence (cf. Höder 2010: 227–225). Two different solutions are attested in HaCOSSA-
The first one is the choice of the active participle, as illustrated in (7):
(7) a. Latin original (Aili 1992: 153)
Numquid ego ideo contempnendus sum, quia mors erat contemptibilis et dura?
surely.not I thus despise-GRND am because death was contemptible and hard
“Should I then be despised because my death was contemptible and hard?”
b. OSw translation (BK 4.40 )
Älla huat ey är jak forsmande for thy at dödhin var smälikin oc hardhir
or whether not am I despise-PTCP.ACT because death-DEF was contemptible and hard
“Or should I be despised because my death was contemptible and hard?”
While this variant might, in certain contexts, be confusing because of the intended passive reading of the
participial form (as opposed to its usual active meaning), it has the advantage of providing a one-word
equivalent to the Latin gerundive, in which not only the morphological structure is similar (a suffix
attached to the verb stem), but even the phonological form resembles the Latin gerundive (a suffix -nd).
Moreover, the distinction between passive gerundival and active participial forms might not have been
perceived as transparent in Medieval Latin either, where an uninflected verbal form ending in -ndo,
homophonous with some inflectional forms of the gerundive, is often used as a morphological variant of
active participles (Stotz 1998: 410–411 [§ 111.26]).
The second translational variant is a morphologically more complex construction consisting of the
active infinitive and a participial form of the modal verb skula ‘shall’, written as two separate graphic
words or univerbated as in (8) (forsma+skulande):
(8) a. Latin original (Aili 1992: 154)
Ideo nullus est contempnendus.
thus nobody is despise-GRND
“therefore nobody is to be despised”
b. OSw translation (BK 4.41 )
thy är ängin forsmaskulande
thus is nobody despise-INF-shall-PTCP
“therefore nobody is to be despised”
In contrast to the use of the bare participle, this construction explicitly expresses the sense of obligation
by incorporating a form of the modal skula, resulting in a more transparent rendering of the Latin
meaning at the expense of morphological isomorphism.
While it is obvious – and even more so than with pronominal RC marking – that the Late OSw
gerundive forms originate in multilingual practices related to the use of both Latin and Swedish in
bilingual groups, and while the two variants occur frequently enough so as to rule out an individual ad
hoc solution to a particular translation problem, the OSw gerundive remains a peripheral affair. In
HaCOSSa-late, it is almost exclusively to be found in translated texts, rarely in OSw originals, and has a
very low overall frequency (in total, less than 30 occurrences in HaCOSSA-late).
Consequently, gerundives did not develop any further and did not make it into any variety or stage of
(written) Swedish that was not characterized by pervasive bilingualism within the speaker group; none of
the variants are found in Present-Day Swedish.
4 Conclusion: monolingual history revisited
The article started out by saying that the history of a language is different from the history of its speech
communities or their communicative practices, and that a ‘linguocentric’ approach to language history can
be rather misguided. Departing from a historical sociolinguistic and language ecological point of view,
the discussion as well as the case studies demonstrated that – given the right circumstances – multilingual
practices of the (remote) past can permeate the language system to such an extent that they are still
reflected by monolingual structures at a much later stage. The history of written Swedish, i.e. the
emergence of specific written registers, begins with its Ausbau in the Late Middle Ages, a development
that was initiated, supported and essentially carried through by a rather small, but enormously influential,
multilingual elite, whose multilingual practices had a formative influence not only on their own text
production and that of their contemporaries, but also on the later development of the language. At least as
far as the history of Standard Swedish is concerned, language contact, multilingualism, and multilingual
practices have been part of it from the outset: There is no such thing as a monolingual history of Swedish,
and even less so Old Swedish.
Aili, Hans (ed.). 1992. Sancta Birgitta. Revelaciones. Book IV (Samlingar utgivna av Svenska fornskriftsällskapet. Serie
2: Latinska skrifter 7.4). Stockholm: Svenska fornskriftsällskapet; Kungl. Vitterhets Historie och Antikvitets Akademien.
Åström, Patrik. 1993. Källorna till den fornsvenska språkhistorien. En översikt över det bevarade bokbeståndet. In Lars Wollin
(ed.), Studier i svensk språkhistoria 3. Förhandlingar vid Tredje sammankomsten för svenska språkets historia, Uppsala, 15–
17 oktober 1992 (Skrifter utgivna av Institutionen för nordiska språk vid Uppsala universitet 34), 229–243. Uppsala:
Institutionen för nordiska språk vid Uppsala universitet.
Blomqvist, Carl Oliver. 2017. Flerspråkighet eller språkförbistring? Finska segment i svenska medeltidsbrev 1350–1526 (Skrifter
utgivna av Institutionen för nordiska språk vid Uppsala universitet 100). Uppsala: Uppsala universitet.
Braunmüller, Kurt. 2002. Language contact during the Old Nordic period I: with the British Isles, Frisia and the Hanseatic
League. In Oskar Bandle et al. (eds.), The Nordic languages. An international handbook of the history of the North Germanic
languages (Handbooks of Linguistics and Communication Science 22), vol. 1, 1028–1039. Berlin/New York: de Gruyter.
Braunmüller, Kurt. 2004. Niederdeutsch und Hochdeutsch im Kontakt mit den skandinavischen Sprachen. In Horst Haider
Munske (ed.), Deutsch im Kontakt mit germanischen Sprachen (Reihe Germanistische Linguistik 248), 1–30. Tübingen:
Braunmüller, Kurt. 2005. Language contacts in the Late Middle Ages and in Early Modern Times. In Oskar Bandle et al. (eds.),
The Nordic languages. An international handbook of the history of the North Germanic languages (Handbooks of Linguistics
and Communication Science 22), vol. 2, 1222–1233. Berlin/New York: de Gruyter.
Comrie, Bernard & Tania Kuteva. 2013a. Relativization on obliques. In Matthew S. Dryer & Martin Haspelmath (eds.), The
world atlas of language structures online, Chapter 123. Leipzig: Max Planck Institute for Evolutionary Anthropology
Comrie, Bernard & Tania Kuteva. 2013b. Relativization on subjects. In Matthew S. Dryer & Martin Haspelmath (eds.), The
world atlas of language structures online, Chapter 122. Leipzig: Max Planck Institute for Evolutionary Anthropology
Comrie, Bernard. 1998. Rethinking the typology of relative clauses. Language Design 1, 59–86.
Deumert, Ana & Wim Vandenbussche. 2003. Research directions in the study of language standardization. In Ana Deumert &
Wim Vandenbussche (eds.), Germanic standardizations. Past to present (Impact. Studies in Language and Society 18), 455–
469. Amsterdam/Philadelphia: Benjamins.
Eliasson, Stig. 2015. The birth of language ecology. Interdisciplinary influences in Einar Haugen’s “The ecology of language”.
Language Sciences 50, 78–92.
Faarlund, Jan Terje. 2004. The syntax of Old Norse. With a survey of the inflectional morphology and a complete bibliography.
Oxford etc.: Oxford University Press.
Gardner-Chloros, Penelope. 2018. Historical and modern studies of code-switching: A tale of mutual enrichment. In Päivi Pahta,
Janne Skaffari & Laura Wright (eds.), Multilingual practices in language history. English and beyond (Language Contact and
Bilingualism 15), 19–36. Boston/Berlin: De Gruyter Mouton.
Gejrot, Claes (ed.). 1988. Diarium Vadstenense. The memorial book of Vadstena Abbey. A critical edition with an introduction
(Acta Universitatis Stockholmiensis. Stockholmer germanistische Forschungen 33). Stockholm: Almqvist & Wiksell.
Gejrot, Claes. 2011. Medeltiden i dataåldern. Om svenskt diplomatarium idag. Kungl. Vitterhets Historie och Antikvitets
Akademiens årsbok 2011, 97–108.
HaCOSSA = Höder, Steffen (ed.). 2011. The Hamburg Corpus of Old Swedish with Syntactic Annotation (HaCOSSA). Version
1.0. Hamburg: Hamburger Zentrum für Sprachkorpora [PID: hdl.handle.net/11022/0000-0000-9D16-7].
Haugen, Einar. 1971. The ecology of language. The Linguistic Reporter 13, Supplement 25, 19−26.
Höder, Steffen. 2010. Sprachausbau im Sprachkontakt. Syntaktischer Wandel im Altschwedischen (Germanistische Bibliothek
35). Heidelberg: Winter.
Höder, Steffen. 2011. Phrases and Clauses Tagging Manual for syntactic analyses of Old Nordic texts encoded as Menotic XML
documents (PaCMan). Version 2.0. Hamburg: University of Hamburg [PID: hdl.handle.net/11022/0000-0000-9D16-7].
Höder, Steffen. 2012. Multilingual constructions: a diasystematic approach to common structures. In Kurt Braunmüller &
Christoph Gabriel (eds.), Multilingual individuals and multilingual societies (Hamburg Studies on Multilingualism 13), 241–
257. Amsterdam/Philadelphia: Benjamins.
Höder, Steffen. 2018. Schwedisch ist gut, Latein ist besser? Spätmittelalterliche Sprachmischung aus normativer Perspektive. In
Thorsten Burkard & Markus Hundt (eds.), Sprachmischung – Mischsprachen: Vom Nutzen und Nachteil gegenseitiger
Sprachbeeinflussung (Kieler Forschungen zur Sprachwissenschaft 9), 145–164. Frankfurt am Main: Lang.
Kloss, Heinz. 1967. ‘Abstand languages’ and ‘Ausbau languages’. Anthropological Linguistics 9(7), 29–41.
Kloss, Heinz. 1978. Die Entwicklung neuer germanischer Kultursprachen seit 1800 (Sprache der Gegenwart 37), 2., erw. Aufl.
Düsseldorf: Schwann [Munich 1952].
Kranich, Svenja, Viktor Becher & Steffen Höder. 2011. A tentative typology of translation-induced language change. In: Svenja
Kranich et al. (eds.), Multilingual discourse production. Diachronic and synchronic perspectives (Hamburg Studies on
Multilingualism 12), 11–43. Amsterdam/Philadelphia: Benjamins.
Langer, Nils & Anna Havinga. 2015. Invisible languages in historical sociolinguistics: a conceptual outline, with examples from
the German-Danish borderlands. In Anna Havinga & Nils Langer (eds.), Invisible languages in the nineteenth century
(Historical Sociolinguistics 2), 1–34. Oxford etc.: Lang.
Le Page, Robert Brock & Andrée Tabouret-Keller. 1985. Acts of identity. Creole-based approaches to language and ethnicity.
Cambridge etc.: Cambridge University Press.
Meyerhoff, Miriam. 2002. Communities of practice. In J.K. Chambers, Peter Trudgill & Natalie Schilling-Estes (eds.), The
handbook of language variation and change, 526–548. Malden/Oxford: Blackwell.
Noreen, Adolf. 1904. Altnordische grammatik: 2: Altschwedische grammatik mit einschluss des altgutnischen (Sammlung kurzer
Grammatiken germanischer Dialekte. A: Hauptreihe 8). Halle: Niemeyer.
Ottosson, Kjartan. 2002. Old Nordic. A definition and delimitation of the period. In Oskar Bandle et al. (eds.), The Nordic
languages. An international handbook of the history of the North Germanic languages (Handbooks of Linguistics and
Communication Science 22), vol. 1, 787–793. Berlin/New York: de Gruyter.
Pahta, Päivi, Janne Skaffari & Laura Wright. 2018. From historical code-switching to multilingual practices in the past. In Päivi
Pahta, Janne Skaffari & Laura Wright (eds.), Multilingual practices in language history. English and beyond (Language
Contact and Bilingualism 15), 3–18. Boston/Berlin: De Gruyter Mouton.
Pettersson, Gertrud. 1976. Om vilken som relativt pronomen. In Nordiska studier i filologi och lingvistik. Festskrift tillägnad
Gösta Holm på 60-årsdagen den 8 juli 1976, 342–352. Lund: Studentlitteratur.
Pettersson, Gertrud. 2005. Svenska språket under sjuhundra år. En historia om svenskan och dess utforskande. 2:a uppl. Lund:
Rajamaa, Ruth. 1992. Systrarnas verksamhet, undervisning och uppfostran i Vadstena kloster 1384–1595. [Stockholm]:
Pedagogiska institutionen, Stockholms universitet.
Rambø, Gro-Renée. 2010. Historiske og sosiale betingelser for språkkontakt mellom nedertysk og skandinavisk i
seinmiddelalderen. Et bidrag til historisk språksosiologi. Oslo: Novus.
SAG = Teleman, Ulf, Staffan Hellberg & Erik Andersson (eds.). 1999. Svenska Akademiens grammatik. 4 vols. [Stockholm]:
Stolt, Birgit. 1964. Die Sprachmischung in Luthers Tischreden. Studien zum Problem der Zweisprachigkeit (Acta Universitatis
Stockholmiensis. Stockholmer germanistische Forschungen 4). Stockholm/Göteborg/Uppsala: Almqvist & Wiksell.
Stotz, Peter. 1998. Handbuch zur lateinischen Sprache des Mittelalters (Handbuch der Altertumswissenschaft 2.5). Vol. 4:
Formenlehre, Syntax und Stilistik. München: Beck.
Teleman, Ulf. 2002. Ära, rikedom och reda. Svensk språkvård och språkpolitik under äldre nyare tid (Skrifter utgivna av Svenska
språknämnden 85). [Stockholm]: Norstedts Ordbok.
Teleman, Ulf. 2003. Swedish. In: Ana Deumert & Wim Vandenbussche (eds.), Germanic standardizations. Past to present
(Impact. Studies in Language and Society 18), 405–429. Amsterdam/Philadelphia: Benjamins.
Timofeeva, Olga. 2013. Of ledenum bocum to engliscum gereorde. Bilingual communities of practice in Anglo-Saxon England.
In Joanna Kopaczyk & Andreas H. Jucker (eds.), Communities of practice in the history of English (Pragmatics & Beyond
New Series 235), 201–224. Amsterdam/Philadelphia: Benjamins.
Wessén, Elias. 1941–1956. Svensk språkhistoria. Vol. 1 (1941): Ljudlära och formlära. Stockholm: Filologiska föreningen vid
Stockholms högskola. Vol. 2 (1943): Ordbildningslära. Stockholm: Filologiska föreningen vid Stockholms högskola. Vol. 3
(1956): Grundlinjer till en historisk syntax (Stockholm studies in Scandinavian philology). Stockholm: Almqvist & Wiksell.
Wiktorsson, Per-Axel (ed.). 1989. Magnus Erikssons landslag enligt Cod. Ups. B 23 (Samlingar utgivna av Svenska
fornskriftsällskapet. Serie 1: Svenska skrifter 78 [issue 258]). Lund: Blom.
Wollin, Lars. 1991a. Ska vi forska om fornsvenska? In Sven-Göran Malmgren & Bo Ralph (eds.), Studier i svensk språkhistoria
2 (Nordistica Gothoburgensia 14), 239–257. Göteborg: Acta Universitatis Gothoburgensis.
Wollin, Lars. 1991b. The monastery of Vadstena. Investigating the great translation workshop in medieval Scandinavia. In Roger
Ellis (ed.), The medieval translator II, 65–88. London: Centre for Medieval Studies, Queen Mary and Westfield College,
University of London.
Wollin, Lars. 2001. När svenskan blev europeisk. Om översättning i svensk medeltid. Folkmålsstudier 40, 117–149.
Wright, Laura. 2002. Code-intermediate phenomena in medieval mixed-language business texts. Language Sciences 24, 471–
This article does not distinguish between bi- and multilingualism, since the difference is not a categorical, but a gradual one.
When the more general term, multilingual(ism), is used, it can also denote bilingualism.
The Hanseatic League had from the 14th century onwards gained enormous economic, political, and cultural influence across
the Baltic Region. In addition to trade, many merchants and craftsmen from Northern Germany even settled in Swedish
towns and formed the basis of a thriving German community there. While many residents of German descent were eventually
absorbed into Swedish society, this was balanced by a constant influx of new immigrants. For an overview, cf. Braunmüller
(2004) and Rambø (2010: 243–310).
“Skulu ok all bref · konunx · laghmanz ok hærazhøfdinga · J thylikum malum · ok andrum · a · suensko skrifuas”, my
translation: “All public charters issued by the king, the lawman and the häradshövdingar [officials on different regional or
local levels] in such and other matters must be written in Swedish” (MELL, Jordabalken 22; Cod. Ups. B 23, fol. 33v;
Wiktorsson 1989: 63).
“En god prædicare forbætre tæssa mørka swænskone æpter latina bokena” (cited in Rajamaa 1992: 248).
TEI, the standard promoted by the Text Encoding Initiative (tei-c.org), is a broad standard for the encoding of all kinds of
texts. Menota is a more specific application of the TEI standard which specializes in medieval Nordic texts; it is promoted by
the Medieval Nordic Text Archive (menota.org). PaCMan stands for “Phrases and Clauses Tagging Manual for syntactic
analyses of Old Nordic texts encoded as Menotic XML documents” and is an annotation scheme that describes the
application of the Menota standard in HaCOSSA. Corpus data can be analysed using XPath queries over both text and
annotations. (For a technical overview, cf. Höder 2011.)
It may be worth pointing out that the annotation of tokens as Latin follows rather restrictive guidelines; proper names, for
instance, are only annotated as Latin if they occur with Latin inflectional suffixes (Höder 2011: 22).
In Early OSw, til took the genitive, but there is a strong tendency towards the accusative throughout the OSw period.
Gerundives are not annotated as such in HaCOSSA, meaning that it is impossible to search for gerundive forms automatically
and get results that are exactly quantifiable. The discussion here is based on an exemplary manual search of Latin gerundives
and OSw translation equivalents in HaCOSSA-late.