More than one language, more than one code?
Linguistic approaches to historical multilingualism
Due to the preoccupation of modern linguistics with speech, the theoretical and data-driven approaches to code-switching until the 2000s were devised on the basis of spoken exchanges.18 A notable exception to this trend as well as a springboard for later work on historical multilingualism can be seen in the early work on mixed-language texts, including the so-called macaronic texts (e.g.
sermons or verse mixing Latin grammar and vernacular words), and on multilingualism in Middle English literature.19 For the study of historical multilingual communication, the analytical framework for code-switching had to be adapted to the medium, and this conceptual and practical work is ongoing.In the collection which lay the groundwork for the current study of historical code-switching, Herbert Schendl and Laura Wright reviewed earlier scholarship and put forward several areas for further linguistic empirical investigation and fine-tuned conceptualisations: exploring a broader range of genres, registers and contexts (linguistic, social, geographical, cultural), creating digital resources and linguistic corpora which do not dismiss the multilingual character of communication and tracing patterns of multilingual language use systematically across time and space.20 These themes have been pursued in business writing,21 medical genres,22 religious writing (sermons in particular),23 in cultural contexts other than the so-far predominant English-focussed perspective24 and in the recent questioning of the monolingual nature of linguistic corpora.25 In search of a broader framework to capture this diversity of contexts, the latest proposal is to extend the outlook on code-switching and talk about multilingual practices26 in general, bringing to the fore the social and individual factors which influence linguistic repertoires.
The complex written code, which is argued for here in the context of medieval urban legal cultures, can certainly be seen as an example of multilingual practice.Legal register as a complex multilingual code
In the past, the multilingual nature of legal culture was the norm, involving Latin and the vernacular.27 Latin was omnipresent and indispensable, but, as intimated earlier, its status as the �higher language’, understood in terms of a typical diglossic situation, should not be taken for granted. To see Latin in a diglossic relationship with the vernacular is to assume that it was authoritative and privileged, while the vernacular served lower-profile roles.28 However, if one takes the situation in late medieval Scotland as an example, two problems arise with adopting this approach. First, the vernacular linguistic fabric was present to varying degrees in legal texts from the beginning of the written record. Place names, personal names29 and terms of legal importance which were lacking from civil and canon law (e.g. kirseth30 or blaa and blodi; see examples below),31 occur in legal Latin as code-switches to the vernacular on the lexical level (see �The level of lexicon’ section). These were not negligible ingredients of legal discourse, which was otherwise overwhelmingly Latin. Quite the contrary, names and other linguistic items of local importance are indexical,32 so in terms of legal significance one could argue that the vernacular could be employed for the most important functions, while Latin was a blueprint, a formulaic construct passed on as a set of practices from one generation of clerks to the next (see �Writers and audiences’ section). Second, in the later middle ages, Scots performed the same functions as Latin – the case study at the end of the chapter illustrates it well – and the difference between the languages was perhaps more one of proportion rather than status.33 The shift towards the vernacular was gradual, but it could only happen because the vernacular had achieved a status on a par with Latin, the traditional written vehicle of law.
Administrative discourse in fifteenth-century Scotland operated within a well-established and fine-grained textual framework34 and utilised larger and smaller formulaic strings to carry the most important meanings.35 This is good evidence for textual standardisation, which indicates a high degree of specialisation and stylistic sophistication of the vernacular in the local legal culture. It is thus problematic to see the relationship between Latin and Scots as strictly hierarchical.In a wider European perspective, important insights on multilingual communication in urban contexts have emerged from the Medieval Urban Literacy project at the University of Utrecht.36 These concern, for example, the shared properties of local customary laws committed to writing and the pivotal role of chanceries in the construction of the written record and the local government. Across Europe, the linguistic fabric of urban administration could include multiple local languages alongside Latin, with Middle Low German asserting itself, for example, as the language of the Hanseatic League.37 In the central European context, the administrative record was mainly kept in Latin, but recent scholarship has argued for the recognition of the role of and mutual complementary relationships between varieties of Polish, Czech, Hungarian and German.38 Scholars working on medieval Romance languages and Latin also see �“multilingualism” [as] more appropriate to the complex linguistic relationships that could be observed in medieval Europe’, as opposed to diglossia.39 Thus, in the context of urban administration, rather than separate languages performing their communicative functions in isolation, we witness a complex code built on the same principles and recognisable across Europe. It was composed of subcodes, Latin and vernacular languages, which complemented each other and drew on each other to carry specific local and global meanings.
The scribes had to be able to access both subcodes simultaneously and switch between them on different levels of linguistic complexity. The audiences who engaged with legal discourse were certainly exposed to its multilingual character, but their engagement with the overall code depended on their level of literacy.Writers and audiences
Administrative records reveal the central role of burgh courts in the day-to-day management of community life in a medieval Scottish town.40 On closer inspection, there existed two communities which exchanged information: the first organised the administrative practices and gave authority to the texts as a reflection and a record of legal events, while the other was constructed by these events – its members recognised themselves through the text (labels, frames of reference, roles captured in the vernacular and in Latin).41 Thus, the notaries, scribes and clerks in a Scottish burgh formed the first community. It can be conceptualised as a �community of practice’,42 which is characterised by mutual engagement, a shared joint enterprise and a particular linguistic repertoire – the familiarity with multilingual legal discourse, its structures and ingredients.43 The form and the content of administrative record had to be accessible not only to the scribes and current participants of the communicative event, but also those who might have wanted to use the record in the future. In fact, more often than not, the intended audience of the legal record was distant in time from the moment of writing, so the linguistic fabric of the record had to be transparent and trustworthy.44 The changeable nature of a living language could have been perceived as a threat to the stability and transparency of the record. As a community of practice, the scribes and notaries overcame this problem by developing, learning, sustaining and transmitting the �way of doing things’ or, in other words, they worked towards standardising legal discourse which would create an authoritative and binding legal record.
The recipients of these texts – the burgesses and other people involved in legal proceedings, including the scribes themselves – can be perceived as the other community, a �text community’.45 It is the text (previously oral, memorised and, presumably, vernacular, now written down in a register-specific code) that becomes a fixed and authoritative point of reference for their roles, obligations and rights in the burgh. The ability to recognise these text-external consequences relied on the ability to interpret a multilingual code. Consider Figure 3.1, which presents a network of translational equivalents for the terms for the most important administrative roles in a Scottish burgh, as captured in the Latin and Scots versions of the Leges Burgorum, the collection of laws pertaining to Scottish royal burghs, compiled in Latin and later in Scots.47
FIGURE 3.1 Administrative roles and term overlaps in Scots (solid outline) and Latin (dashed outline) in Leges Burgorum.46
The terms in Latin and in Scots do not match neatly one-to-one; they create a gradient scale of near-synonyms, shading into each other through reinterpretation in the other language. Yet, the terms define roles which on a given occasion would not be subject to a loose interpretation. The people of the burgh, both the literate multilingual writers and the illiterate (possibly monoglot) audiences, would know what to expect of an aldirman, perhaps more than of a prepositus.48 What this situation suggests is that certain local vernacular practices may not have been easily rendered by a single Latin term, steeped in a more generic legal culture. Conversely, the conventional Latin terms may have been too rigid to capture what a given person actually did in a burgh.49 Nevertheless, as a text community, the people of the burgh had to be able to recognise the meanings embedded in the multilingual documents.
Admittedly, meanings carried through this complex code bear implications which go beyond the active consumers of the administrative record and concern broader communities, if not societies and nations.Switching (sub)codes in medieval legal discourse: a structural approach
Switching codes may happen on all levels of linguistic structure and be visually marked or unmarked on all these levels by means of an array of visual cues.50 In the context of legal discourse, as suggested earlier, one complex code is employed to achieve the communicative goals, which means that any switches between Latin and the vernacular should be conceptualised as switches between subcodes. Before the discussion moves on to a single case study, I illustrate the similarities in multilingual practice in the urban legal cultures of fifteenth-century Scotland and Poland. In both locations, the complex code can incorporate items from different languages on various structural levels, from macro (the manuscript) to micro (spelling choices) levels.
The level of macrogenre
Various genres may function within a larger collection of texts, which in itself becomes a macrogenre.51 Some medieval manuscripts are compiled out of different genres gathered by a scribe, as is the case with Sammelhandschriften,52 or they consist of loose parts bound together by a collector. Miscellanies and medieval cartularies are good examples of such collections.53 Other administrative records may be interpreted as multi-genre, depending on how fine-grained a typology of the forms of legal writing one adopts. In any case, the code-switching may be present on the level of the whole collection, for example, where some genres have Latin as the main language, while some others have the vernacular. In my primary material, the eROThA corpus can be said to have been written wholesale in Latin; it is not the case that one entry is in Latin and an adjacent one is in Old Polish. Switches to the vernacular happen on deeper structural levels. For LAOS it is hard to tell, as the corpus is purposefully made of texts identified as Scots, which may have been part of larger multilingual compilations.
The level of discourse
Thus, going deeper into the textual fabric of a genre, one encounters particular discourse units, for instance, rubrics, incipits and explicits, reported speech passages, lists of names and so on. Switching from Latin to the vernacular for oath-taking is the primary focus of the eROThA digitisation project, whose tagging procedures have enabled tracing the switches in a systematic way.54
Example (1) Latin to Polish, Kalisz Land Court Book 2, f.115, R. 407 (1413), 55
Testes ducit Jutka de Bobri contra Halscam filiam suam: primus Johannes de Cirnino, qui fuit fideiussor pro dotalicio, secundus Staszek Rzekta, tercius Janussius de Cotarbi, quartus Thomas de ibidem, quintus Laurencius de Crosznino, sextus Albertus de ibidem.
Rotha
Jaco czoſm ranczil za hal iutką halſcze
ſzecznadcze grziwen poſſagu to geſt
yeg zaplaczono Ceteri ad testimonium: Jaco
halſcze zaplaczono etc.
Translation: �Witnesses presented by Jutka of Bobry against his daughter, Halszka: first Jan of Czernino, who is a guarantor of the dowry, second Staszek Rzekta, third Janusz of Kotarby, fourth Tomasz thereof, fifth Wawrzyniec of Krosznino, sixth Albert thereof. Oath. That I have guaranteed on Jutka’s behalf for Halszka sixteen marcs in dowry which has been paid to her. Ceteri ad testimonium [Others testify accordingly]: That Halszka was paid etc.’
Figure 3.2 presents the manuscript image of the text in Example (1) to illustrate the point about visual cues for code-switching and changes in discourse units. Notice that the multilingual text is written by the same scribe who chose to introduce line breaks between the Latin preamble, the code-switch and discourse trigger Rotha, and the text of the oath. In addition, on closer inspection, the oath contains a visually unmarked switch to Latin on the level of the clause (ceteri ad testimonium), embedded in the Polish discourse unit.
FIGURE 3.2 Kalisz Land Court Book 2, f.115, R. 407 (1413). Copyright State Archive in Poznań (Archiwum Państwowe w Poznaniu).
The level of syntax
In medieval legal record, Latin and the vernacular can swap within larger discourse units on the level of sentence or even within a sentence. We have already seen this earlier in the Polish oath in Example (1), where the additional testimonies are introduced by an unmediated Latin clause.56 Consider also the switch from Scots to Latin at the level of the clause unit in the following example:
Example (2) Scots to Latin, National Records of Scotland B22/22/2, f. 52v, notarial protocol book (1488), LAOS file #785
…in the maist strait stile And sekerest forme of obligacioune yat57 can be deuisit but fraud or gile Renunciand al preuilege of law Ciuile or Canone actis statutis of parliament or generale counsale maid or to be maid letterez of supersedere Respitt the terme contenit in the law de pecunia non immunata And al vtheris Remedis yat may be proponit in contrare…
Translation: �In the straightest style and most secure form of obligation that can be devised without fraud or deceit, renouncing all privilege of civil or canon law, acts, statutes of parliaments or General Council made or to be made the letters of superseder [an agreement among creditors to postpone action] to delay the term contained in the law de pecunia non immunata. And all other remedies that may be proposed to the contrary...’
The reference to the specific legal provision concerning financial matters is done by means of a switch to Latin (de pecunia…) within a longer Scots passage and back again. In fact, that switch does not contain a self-standing verb (which is implicit), so it can also illustrate the interaction of languages within the clause on a lower structural level.
The level of lexicon
Smaller linguistic units from a different language can also be incorporated into longer passages in the main language. In such cases the issue of how to distinguish between borrowing and code-switching comes to the fore.58 In general, there are several factors to consider: the frequency across the linguistic record for a given period, the inclusion in dictionaries, morphological productivity – or the ability to adapt into the grammatical structure of the main language, e.g. by attaching inflectional endings, and spelling adaptation. The examples below illustrate code-switching on the lexical level – it would be difficult to argue that the switched items were borrowings and had already entered the recipient language and blended in with the rest of its lexicon.
Example (3) Latin to Scots: example from Leges Burgorum, chapter xxxi:59
L: Et est retinendum quod in placitis burgorum utitur Twertnay in defensionibus defendando wrang and unlawe
Sc: And it is to wyt that in borow mutis thar is hantyd and oysyt thuertnay in defendande wrang and unlawe
Translation: �And it is to be known that at the burgh moots [formal meetings] the custom of twertnay [a flat denial] is used in defence if accused of wrongdoings...’
Example (4) Latin to Polish, Gniezno Land Court Book 3, f. 19v, R. 310 (1424), https://rotha.ehum.psnc.pl/breeze/Gn.310
… infra meridiem, vulgariter dopoludnya
Translation: �before noon, in the local language before noon’
It is often the case that one-off word-level switches are flagged in the main language of the passage, as illustrated in (4) where the Latin vulgariter prepares the reader for the switch to the vernacular.
The level of morphology
On this level, we look inside the word structure. There are cases in the medieval records where one linguistic system provides inflections to be attached to lexical items from another linguistic system. This phenomenon can be conceptualised as going even deeper into the levels of linguistic structural complexity – a code-switch on the level of morphology. Consider the Latin inflection (third declension, masculine) on the Polish first name Jaszko in (5).
Example (5) Polish to Latin, Poznań Land Court Book 1, f. 112, R. 164 (1393), https://rotha.ehum.psnc.pl/breeze/Pn.164
Item Pelka tenetur 6 penas non comparentes et 2 pokupna et super suo patre contra Jaszkonem Szoboczszky …
Translation: �Next Pelka has six different fines and two damage fees on his father’s behalf against Jaszko[+Lat.infl] Szoboczszky...’
This is a frequent practice in administrative record; interestingly, last names are not prone to code-switching on this level, cf. Szoboczszky, which remains uninflected.60
Example (6) Scots to Latin (?), National Records of Scotland, MS 34.4.3, f. 77r, arbitration (1437), LAOS file #250
… s(er) Jamys of ogyluy and ye remanent of ye Jug(is) arbitror(is) (con)tenit Jn ye fyrst (com)p(ro)misß…
Translation: �sir James of Ogilvy and the remaining judges arbitrators mentioned in the first settlement’
In this example from a Scots register, the inflectional endings and other parts of words were abbreviated – the traditional expansions of particular abbreviations are provided in parentheses in (6). The question arises whether these symbolic representations of some underlying linguistic content take the Scots grammar as the point of reference, or perhaps imply the switch to Latin, which is ultimately the source of the abbreviating system in Scots. Such ambiguous forms, copiously encountered in Anglo-Norman business writing, have been labelled visual diamorphs.61 These can potentially be code-switches on the level of a morpheme, marked visually by means of an ambiguous symbol.
The level of orthography
A respelling of a word in a different orthographic system can also be seen as code-switching, this time on the level of spelling. For instance, in order to fit it into a Latin administrative record, a person’s name may be respelled according to Latin conventions, regardless of having its own written form in the vernacular. In a way, the word stays in the vernacular (and would be read out as such, if required) but its representation on the page is rendered through another language’s spelling conventions. To use an example from the Polish land courts, the same Polish diminutive name appears in the record as Jasek, Jaszko (see Example 5) or Jasco, but the last version is a Latin respelling to fit it into the surrounding context (7).
Example (7) Poznań Land Court Cook 1, f. 185, R. 339 (1397), https://rotha.ehum.psnc.pl/breeze/Pn.339
Jaco Jaſek przi jal […] y wſøl mego cloueka y ſzelaſa Item Jasco iurabit: Czo …
Translation: �That Jaszek arrived […] and captured my man and my iron plough Next Jasco will testify: As …’
In the Scots material, the opposite direction of an orthographic switch can be discerned. Some scribes respelled the Latin formula confirming the notary’s identity, in manu propria, according to the Scots spelling convention. In Older Scots, the sequence can stand for the vowel /u/, as in blue spelled or virtue spelled, so using this convention within a Latin word is an instance of code-switching to the vernacular on the level of spelling, see (8) and (9).
Example (8) National Library of Scotland CH 1494, Spottiswoode Papers (1490), obligation, LAOS file #189
The ʒere of god Jm iiijc and Nyntj ʒer(is) alexand(er) fulfurt MANEW p(ro)p(ri)a
Translation: �The year of God 1490 years Alexander Fulfurt in his own hand’
Example (9) National Records of Scotland, B20/10/1, p.52.
Ye quhylk day it is funde be ane inquest yt a(n)d(er)o fowlar burges(s) of du(n)fermlyne Js nerest & lachfull ayr to s(s) Jhon terwat hys eme wmquhyll cwret off Du(n)fermlyne
Archybalde stewart baylʒe manw p(ro)p(ri)a
Translation: �The which day it is found by an inquest that Andrew Fowler burgess of Dunfermline is the nearest and lawful heir to sir John Terwart his uncle, the late curate of Dunfermline, Archibald Stewart baillie in his own hand’
It is also possible to mark the orthographic switch visually by the change of hand or script,62 which implies awareness of the change between linguistic systems on the part of the scribe. In the examples listed earlier, there was no visible change in the script when switching to the spelling conventions of the other language.
Visual aspects of code-switching on the page
The earlier discussion has already highlighted several cases where code-switching was marked visually on the page. Visual cues may be very diverse: ink/colour change, letter size, line breaks, the choice to use the margin, orientation, hands, pointers, and so on. Various typologies have been suggested within the field of manuscript studies and the �pragmatics of the page’.63 In such cases, the interaction of the two languages, the two (sub)codes making up the complex multilingual legal code, is made explicit and brought to the fore. The function of visual marking should be best considered on a case-by-case basis but it seems to be related to the awareness of the scribe and to the processing capabilities of the audience, who might welcome the convenience of a visual prompt to decode the message. At the same time, the absence of visual cues for code-switching, quite prominent in the material consulted, indicates a closely integrated linguistic repertoire – a complex multilingual code – which draws on the available languages without drawing attention to the ingredients.