Vernacularisation in the Aberdeen Council Registers: a quantitative analysis
The corpus and method
The Law in the Aberdeen Council Registers (LACR) project team fully transcribed volumes one to eight of the registers (excluding the material written between 1414 and 1433, which is missing).
This forms the publication Aberdeen Registers Online: 1398–1511 (ARO).16 The results presented here are based upon an examination of the corpus as it stood during its period of creation. At the time of study, the majority of the initial transcriptions, which can be described as semi-diplomatic, were edited and their TEI-compliant annotations checked.17 Table 4.1 specifies the size of the overall corpus and the individual volumes. At the time of study, the corpus contained 34,077 entries, which made up 1,805,670 tokens.18 Not every volume covers the same number of administrative years. The average number of entries and tokens per administrative year in Table 4.1 reveals an overall increase in how much was written down over the years, with the highest average in volume eight (518 entries and 33,106 tokens per administrative year). This may either indicate that recording burgh business became increasingly important or that more issues arose in the burgh that needed to be recorded.TABLE 4.1
The Aberdeen Council Registers corpus
Volume
Time period
Number of entries
Number of tokens
Average number of entries per admin. year
Average number of tokens per admin. yeara
Vol. 1
1398–1407
2,528
86,326
316
10,791
Vol. 2
1408–1414
1,490
44,611
248
7,435
Vol. 4
1433–1448
3,755
150,996
250
10,066
Vol. 5.1
1448–1468
4,418
207,601
210
9,886
Vol. 5.2
1441–1471
1,052
44,157
38
1,577
Vol. 6
1466–1486
9,047
556,506
431
26,500
Vol.
71487–1501
6,606
384,411
440
25,627
Vol. 8
1501–1511
5,181
331,062
518
33,106
Total
1398–1511
34,077
1,805,670
a The administrative years 1403/1404 and 1404/1405 are not recorded in volume one. The few entries that were dated 1405 (without a specific date) are here included in the administrative year 1405/1406.
As Table 4.1 indicates, volume five part two (5.2) overlaps with volumes four, five part one (5.1) and six. This is because volume 5.2 contains the guild court records, which were separated from other burgh business between 1441 and 1472.19 Initially, all volumes were treated as separate entities in the quantitative analyses, in order to detect any differences in language use between the guild court and other burgh business. In the final calculations, the figures of this overlapping period were, however, combined in order to trace the vernacularisation process consistently across the volumes (see sections �Quantitative analysis of the matrix language of entries’, �Quantitative analysis of the number of words in Scots as opposed to Latin’ and �Comparisons and explanations’).
The quantitative analyses were carried out on the prototype of the LACR online search platform, which was only accessible to the research team at the time of writing. The platform, which is now freely available online, serves as a repository for the images of the original documents and the ARO transcriptions.20 The search platform allows users to query the transcriptions in various ways.21 Search terms can be specified by time periods, volumes, language and other factors.
For the analyses presented here, XQuery was used to quantify the number of entries with Scots and with Latin as the matrix language as well as the number of words in Scots or Latin. The �matrix language’ is here defined as the dominant language of an entry.22 This generally means that if the majority of words in an entry were Scots, then the entry was annotated as Scots.23 The initial annotations applied by the transcribers were checked automatically by a tool developed by Wim Peters.
This tool matched all words in each entry to a lookup list of Latin words compiled from Latin entries in the Aberdeen Council Registers. Using a probability threshold of 0.5, any entry that was marked up as Latin but with the number of Latin words being below this threshold was flagged for the editor, Edda Frankot, to review. This tool thus guaranteed the consistent assignment of languages to individual entries. In most cases, the matrix language was easy to determine. For some entries, however, it was not possible to determine a matrix language based on the �dominance’ approach (particularly in lists of names with Latinised first names but with Scots surnames). In these instances, the attribute “mul” for �multiple languages’ was used in the language tags.The XQuery searches were divided into administrative years, running from the first Monday after Michaelmas (29 September) to the following year. The administrative years were determined manually by going through the dates of individual entries in the transcriptions and noting the entry IDs of the first and last entries of each administrative year. One issue here is that the entries are not always in diachronic order, particularly between 1466 and 1472, the period covered at the beginning of volume six. In this particular case, the entries were listed in a time period (1466–1472), rather than individual administrative years. There are also copies of letters, dated with their original date, that were incorporated into an entry under the date of the entry rather than when the original letter was written. The dates of these original letters were not taken into account in the quantitative analyses described below.
After determining the administrative years manually, XQuery was used to determine the number of entries with Scots/Latin/Dutch/multiple languages as the matrix language (see Appendix, XQuery 1). The query provided in the appendix finds and lists all entries annotated as Scots for the administrative year 1433/1434, which spans from the first entry on page 1 of volume four to (and including) the first entry on page 23 of volume four.
The same query was then used for Latin, Dutch and entries with multiple languages by replacing the language attribute “sco” with “lat”, “nld” and “mul”, respectively, and all results were recorded in a spreadsheet. The entry identification numbers (xml:id) were then changed for the specific administrative years and the query was run again. To count tokens in Scots, Latin, Dutch and multiple languages, a more complex query was used (see Appendix, XQuery 2). This query tokenised words annotated as Scots, Latin, Dutch and multiple languages per administrative year, which was again expressed by the range of xml:id (in the query provided in the appendix it is 1433/1434), in monolingual as well as multilingual entries. In other words, it did not just count the words in each Scots entry but also the Scots words within Latin entries. Words were defined as a string of letters or numbers (0–9, a–z, A–Z), so that punctuation marks were not counted as separate words.It has to be pointed out that these queries rely on the quality of annotations. The editing process carried out by the project team during and after the research presented here may result in slight differences in the findings if the same analyses were carried out again. Furthermore, it is not always possible to identify a word as Scots or Latin due to abbreviations. The abbreviation barrell’, for example, could stand for a Latin or Scots word, and the ampersand, which was expanded silently in the transcriptions, could stand for et or and. It is also problematic to determine when a borrowed Latin word has become a naturalised part of Scots. This distinction between borrowing and code-switching, the latter defined here as �the use of more than one language in the course of a single communicative episode’,24 is a much-discussed topic in linguistic research.25 Matras points out that these two phenomena can either be considered as two distinct phenomena or, and this is the view taken in this chapter, as �related points on a continuum’.26 For example, the word item, one of the most frequently used words in the corpus, could be categorised as Latin or it could be seen as a word that had become Scots by the fifteenth century.
The Dictionary of the Scots Language lists examples of item being used in Scottish documents from 1397, suggesting that item can be considered a naturalised Scots word in fifteenth-century records.27 However, even attestations in dictionaries are not always a reliable source for determining whether a word has become naturalised, and therefore constitutes borrowing rather than single-word code-switching.28 Due to these issues, Trotter urges researchers to take a multilingual approach when working on late medieval Britain.29 During this time, �for the educated at least, monolingualism was the exception and not the norm’.30 Working with XML annotations does not make it easy to take such a multilingual approach as it forces the transcriber to categorise entries and words into certain languages. This does, however, not mean that the same word can only be read as a Latin or Scots word. Returning to item, this word was annotated as Latin when occurring in a Latin context, i.e. when the preceding and/or following words were clearly Latin, and as Scots in a Scots context.While the categorisation of individual words into specific languages remains problematic, the matrix languages of entries in the Aberdeen Council Registers are in the majority of cases clear. The quantitative analysis of the matrix language of entries and the use of the �multiple languages’ attribute surpass the problem of categorising individual words to some extent. However, an analysis of the matrix language ignores instances of language mixing and code-switching within individual entries. The �Code-switching in the Aberdeen Council Registers’ section is dedicated to some of these instances. Furthermore, a quantitative analysis of the words in Scots as opposed to Latin was carried out to illustrate that Scots did occur in administrative years where no entries with Scots as the matrix language were found (see year 1438/1439). Although specific numbers are presented in the �Quantitative analysis of the number of words in Scots as opposed to Latin’ section, the figures have to be seen as approximations due to the issues of annotating words in a specific language described earlier.
Quantitative analysis of the matrix language of entries
In this section, the quantitative analysis of the matrix language, i.e. the dominant language, of entries is presented, before moving on to the number of words in Scots as opposed to Latin in the following section. The findings of these analyses are then compared in the �Comparisons and explanations’ section.
TABLE 4.2
Matrix languages of entries in the Aberdeen Council Registers (1398–1511)
Volume
Time period
Number of entries
Scots
Latin
Dutch
Multiple
n
%
n
%
n
%
n
%
Vol. 1
1398–1407
2,528
6
0.24
2,508
99.21
0
0
14
0.55
Vol. 2
1408–1414
1,490
9
0.60
1,467
98.46
0
0
14
0.94
Vol. 4
1433–1448
3,755
209
5.57
3,540
94.27
0
0
6
0.16
Vol. 5.1
1448–1468
4,418
317
7.18
4,092
92.62
0
0
9
0.20
Vol. 5.2
1441–1471
1,052
226
21.48
823
78.23
1
0.10
2
0.19
Vol. 6
1466–1486
9,047
2,958
32.70
6,087
67.28
1
0.01
1
0.01
Vol. 7
1487–1501
6,606
3,601
54.51
3,004
45.47
0
0
1
0.02
Vol. 8
1501–1511
5,181
3,310
63.89
1,870
36.09
0
0
1
0.02
As Table 4.2 shows, Scots and Latin are not the only languages used in the Aberdeen Council Registers. In volume 5.2, a letter in Middle Dutch from Antwerp is copied into the council registers (see entry ARO-5-0714-02), with another, albeit very short entry in Middle Dutch, in volume six (entry ARO-6-0691-04). These two entries point to the close trading relationships between Aberdeen and the Low Countries, which cannot be discussed in more detail here.31 Table 4.2 also provides the number of entries with �multiple languages’, i.e. entries without a �dominant’ language. One of these entries will be discussed further in the �Code-switching in the Aberdeen Council Registers’ section. This section will instead focus on the vernacularisation in volumes one to eight of the Aberdeen Council Registers.
The quantitative analysis of the council registers revealed a considerable diachronic increase in the number of entries with Scots as the matrix language (see Table 4.2 and Figure 4.1). The number of entries with Scots as the matrix language remains under 1 per cent in the first two volumes (six entries in volume one and nine entries in volume two). It can be assumed but not proven that this percentage is higher in the material written between 1414 and 1433, which is missing. Given that merely 5.6 per cent of entries, i.e. 209 entries, had Scots as the matrix language in volume four and 7.2 per cent in volume 5.1 (317 entries), it is unlikely that the percentage of Scots entries in the material written between 1414 and 1433 is higher than 7 per cent. While this needs to remain as speculation, a clear increase in the use of Scots can be seen between volumes four and eight, with Scots gradually becoming the more dominant language at the expense of Latin, which decreases from 94.3 per cent in volume four to 36.1 per cent in volume eight. Of interest is also the difference between volume 5.1, in which 7.2 per cent of entries have Scots as the matrix language, and volume 5.2, in which considerably more entries with Scots as the matrix language can be found, namely 21.5 per cent. Given that volume 5.2 deals with the guild court business, this difference indicates that the content of the entries was one of the factors for language choice, with Scots being used more frequently for guild court business than for other burgh business recorded in volume 5.1. The entries in volume 5.2 are, however, not as regular as the ones in 5.1, which leads to divergent percentages in individual administrative years in volume 5.2. For example, in the year 1460/1461, only one entry was recorded. This entry was written in Latin, which means that in this particular year, 100 per cent of entries had Latin as the matrix language. Due to these discrepancies, the calculations of volume 5.2 were incorporated into the numbers from volumes four, five and six for the individual administrative years. The diachronic development across volumes four to eight, divided into the administrative years, is shown in Figure 4.1.32
FIGURE 4.1 Percentage of entries with Scots as the matrix language in the Aberdeen Council Registers (1433–1511).
The dotted polynomial trend line in Figure 4.1 illustrates a relatively gradual increase in the use of Scots between 1433 and 1511. The continuous black line indicates, however, that there is a particular strong increase of Scots in 1443/1444, in which the entries with Scots as the matrix language reach over the 10 per cent mark for the first time (16.2 per cent, i.e. 59 entries, in 1443/1444 and 18.1 per cent, i.e. 95 entries, in 1444/1445). The next time such a high percentage of entries with Scots as the matrix language is reached is 20 years later, in 1464/1465 (16.0 per cent, i.e. 45 entries). From 1466/1467 onwards, the percentage of entries with Scots as the matrix language remains above 10 per cent, with further increases to 25.7 per cent in 1472/1473 (211 entries) and to 31.2 per cent in 1474/1475 (197 entries). Further considerable increases can then be detected between 1481/1482 and 1484/1485, the latter year being the one with the highest percentage of entries with Scots as the matrix language in the fifteenth century. During these four administrative years, the percentage increased from 36.5 per cent in 1481/1482 to 64.2 per cent in 1484/1485. The year before that constitutes the first year where Scots entries outnumber Latin ones, with 57.5 per cent of entries having Scots as the matrix language. After the peak in 1484/1485, a relatively balanced use of Scots and Latin can be observed, when considering the matrix language of entries. A further increase follows in the sixteenth century, during which Scots is consistently used in more entries than Latin. The highest percentage of entries with Scots as the dominant language can be found in 1506/1507 (70.3 per cent). After that, the percentage drops below the 70 per cent mark again. We can, therefore, conclude that the Aberdeen Council Registers were not fully vernacularised by 1511, with Latin remaining to be used as the matrix language in a considerable number of entries. The following section will report on the number of Scots as opposed to Latin words in the same period of time, before providing explanations for the increasing use of Scots in the council registers.
Quantitative analysis of the number of words in Scots as opposed to Latin
It is, of course, no surprise that the increase in the number of words in Scots runs parallel to the increase of entries with Scots as the matrix language. However, the percentages of Scots words are higher than the percentages of Scots entries in each of the volumes. This can be explained by the length of individual entries with some Latin entries (such as the admission of the burgesses) being very short and the use of Scots within entries with Latin as the matrix language (see the �Code-switching in the Aberdeen Council Registers’ section).
As Table 4.3 shows, the percentage of Scots words remains equally low in volumes one and two (1.3 per cent). Entries with multiple languages are, however, more common in volume two. Rather than categorising individual words into Latin and Scots in these entries, the words were annotated as �multiple’, with 7.6 per cent of �multiple language words’ in volume two and 2.5 per cent in volume one. This lowered the percentage of Latin words from 96.2 per cent in volume one to 91.1 per cent in volume 2. In volume four, 9.3 per cent of tokens, i.e. 14,072 words, were annotated as Scots. This percentage increases to 20.2 per cent (41,967 tokens) in volume 5.1 and to 40.9 per cent (18,068 tokens) in volume 5.2. In volume six, slightly more words are annotated as Scots (283,108 tokens, 50.9 per cent) than as Latin (273,364 tokens, 49.1 per cent) and in volume seven, Scots is the dominant language, with 260,758 tokens, which equates to 67.8 per cent. A further increase in the percentage of Scots words can be seen in volume eight, with 71.4 per cent of words being annotated as Scots.
TABLE 4.3
Number of tokens in different languages in the Aberdeen Council Registers (1398–1511)
Volume
Time period
Number of tokens
Scots
Latin
Dutch
Multiple
n
%
n
%
n
%
n
%
Vol. 1
1398–1407
86,326
1,122
1.30
83,064
96.22
0
0
2,140
2.48
Vol. 2
1408–1414
44,611
585
1.31
40,656
91.13
0
0
3,370
7.55
Vol. 4
1433–1448
150,996
14,072
9.32
135,267
89.58
0
0
1,657
1.10
Vol. 5.1
1448–1468
207,601
41,967
20.22
157,582
75.91
0
0
8,052
3.88
Vol. 5.2
1441–1471
44,157
18,068
40.92
25,843
58.53
89
0.20
157
0.36
Vol. 6
1466–1486
556,506
283,108
50.87
273,364
49.12
18
0.00
16
0.00
Vol. 7
1487–1501
384,411
260,758
67.83
123,617
32.16
0
0
36
0.01
Vol. 8
1501–1511
331,062
236,387
71.40
94,550
28.56
3
0.00
122
0.04
Figure 4.2 illustrates the increase of tokens annotated as Scots between 1433 and 1511. While specific numbers are presented in Table 4.3 and were used as the basis for the graph below, these numbers do not fully account for the fact that it is not always possible to strictly categorise words as Latin or Scots, even if some words are marked as appearing in entries with �multiple languages’. For example, the word �item’ (see earlier point), used at the beginning of many entries, was annotated as Latin when the matrix language of the rest of the entry was Latin but as Scots when the matrix language was Scots. The fuzziness of the line graph below highlights this issue, with exact numbers being difficult if not impossible to determine. This graph merely approximates the number of tokens that can be classified as �Scots’ and the increasing vernacularisation over time. In order to avoid evoking an unrealistic image of an entirely clear distinction between Latin and Scots, the Latin numbers are not represented in the graph. As Trotter points out:
FIGURE 4.2 Percentage of tokens annotated as Scots in the Aberdeen Council Registers (1433–1511).
[n]either for the modern lexicographer nor for the user of the languages of Britain in the middle ages can clear-cut distinctions between vernaculars, or between vernacular and Latin, be considered an accurate representation of linguistic and social reality. Glossators and writers alike show scant regard for watertight divisions between languages.33
The fact that the Aberdeen Council Registers are multilingual in nature needs to be kept in mind and is discussed in more detail in the �Code-switching in the Aberdeen Council Registers’ section.
Despite the issues surrounding the categorisation of individual words as Scots, a clear increase in the use of Scots can be observed in the period under investigation (see polynomial trend line in Figure 4.2). With the exception of the first administrative year (1433–1434), the number of words annotated as Scots remains below 10 per cent until 1441/1442. After that, a drastic increase of Scots can be observed between 1441/1442 and 1445/1446, with the percentage of words annotated as Scots rising from approximately 10 per cent to over 25 per cent. In 1464/1465, more than every third word can be considered Scots, with further increases in the 1470s. The year 1474/1475 is the first in which more words are annotated as Scots (58.5 per cent) than as Latin (41.5 per cent). Scots can be considered the dominant language in the Aberdeen Council Registers from the early 1480s onwards, with percentages of Scots words ranging from 53 per cent in 1481/1482 to almost 75 per cent in 1485/1486 in the 1480s. The highest percentage of words annotated as Scots can be found in 1508/1509 (79.3 per cent). However, as pointed out in the previous section, Latin clearly remains to be used throughout the period under investigation, indicating its continuous importance in legal writing.
Despite the problematic issue of categorising words into Latin or Scots, it is worth quantifying the number of Scots words as it allows us to uncover Scots words used within Latin entries and thus complements the quantitative analysis of the matrix language of entries. In the following section, the differences between the results of these analyses are discussed, before providing some explanations for the vernacularisation developments in the council registers.
Comparisons and explanations
In order to show vernacularisation processes in the Aberdeen Council Registers more clearly, the period covered by volumes four to eight was divided into 16 intervals of five administrative years each (see Figure 4.3).34 In the first two of these intervals (1433/1434–1442/1443), the percentage of Scots entries and words remains below the 10 per cent mark. After that, the numbers for Scots words diverge from the numbers for entries with Scots as the matrix language. In the third interval (1443/1444–1447/1448), there is a difference of almost 10 per cent between them (Scots entries: 13.1 per cent, Scots words: 22.3 per cent). After the rather drastic increase in the use of Scots between the first two and the third interval, the process continues more gradually between 1448/1449 and 1481/1482, with a difference between the percentage of Scots entries and Scots words ranging from 11.4 per cent in 1448/1449–1452/1453 to 22.8 per cent in 1472/1473–1476/1477. After another considerable increase between 1481/1482 and 1486/1487, the number of Scots entries and words remains fairly stable between 1487/1488 and 1501/1502. A further increase in the use of Scots can be seen in the final two intervals, with the percentage of Scots words staying above the percentage of Scots entries. A comparison of the two graphs plotted in Figure 4.3 illustrates that solely counting the matrix language of entries or the dominant language of texts (which seems to be the preferred research method currently employed) does not provide a full and accurate picture of vernacularisation processes. While the categorisation of individual words into distinct languages remains problematic, the method employed here highlights the fact that the vernacular may have been used more widely earlier on than previously thought.
The particularly drastic increase between 1481 and 1482, when 36.5 per cent of entries and 53 per cent of words were annotated as Scots, and the following three administrative years35 partly coincides with the beginning of the so-called Sasine Register of Aberdeen (1484–1502), one of the only Scottish burgh protocol books surviving from the fifteenth century.36 Frankot explains that the misleading term �sasine’, which refers to the transfer of ownership of property, was added later. In fact, this register includes entries relating to the sheriff court in Scotland, copies of legal documents and various legal entries, many but not all of which concern property. Frankot speculates that some of the business recorded in Latin in the Aberdeen Council Registers was transferred to the Sasine Registers in 1484, which would explain the relative drastic decrease of Latin in that year. The Sasine Register itself is mostly written in Latin.37
FIGURE 4.3 Diachronic development of the use of Scots in the Aberdeen Council Registers in per cent (1433–1511).
The fact that Scots became the more dominant language in the records of parliament after 1425 (see �Introduction’) and the fact that the vernacular was used almost exclusively in other Scottish burgh records, such as the Burgh Court Book (1457–1479) from Newburgh in Fife, by 148038 points to a more local rather than a national development in the vernacularisation of the Aberdeen Council Registers. This raises questions about the spread of the vernacular in legal records of different Scottish burghs, which cannot be addressed here. It also highlights the point made in the �Introduction’: vernacularisation processes happened at different times in different regions and in different records, making it difficult to determine a general timeline of vernacularisation. It can, however, be generally assumed that all legal records in Scotland were multilingual in the fifteenth century, with code-switching and language mixing occurring frequently. The following section will discuss the multilingual nature of the Aberdeen Council Registers, focussing on instances of code-switching, before moving on to more general conclusions.
Code-switching in the Aberdeen Council Registers
As the earlier analyses show, Scots and Latin are used in each of the first eight volumes of the Aberdeen Council Registers. In the TEI-compliant transcriptions of the Aberdeen Council Registers, Scots words within Latin entries and Latin words within Scots entries were annotated with the tag, which makes it easy to quantify the number of code-switches (CS) into Latin and Scots in the records. As mentioned earlier, code-switching is here defined as �the use of more than one language in the course of a single communicative episode’.39 Table 4.4 lists the number of these CS, separated into volumes. The normalised frequency of these CS (Scots and Latin together) is also provided to compare the relative frequency of the CS per 1,000 entries in the eight volumes (as explained earlier, relying on data compiled at the time of study).
TABLE 4.4
Number of code-switches in the Aberdeen Council Registers (1398–1511)
Volume
Time period
Number of entries
Number of tokens
CS into Scots
CS into Latin
Total number of CS
Normalised CS per 1,000 entries
% of entries annotated as �Scots’
Vol. 1
1398–1407
2,528
86,326
187
28
215
85.05
0.24
Vol. 2
1408–1414
1,490
44,611
92
1
93
62.42
0.60
Vol. 4
1433–1448
3,755
150,996
307
34
341
90.81
5.57
Vol. 5.1
1448–1468
4,418
207,601
402
123
525
118.83
7.18
Vol. 5.2
1441–1471
1,052
44,157
88
46
134
127.38
21.58
Vol. 6
1466–1486
9,047
556,506
325
283
608
67.20
32.70
Vol. 7
1487–1501
6,606
384,411
167
376
543
82.20
54.51
Vol. 8
1501–1511
5,181
331,062
135
279
414
79.91
63.89
Schendl comments on the chronological correlation between vernacularisation and the use of code-switching, but concedes that further research into this correlation is necessary.40 Some conclusions concerning the link between code-switching and vernacularisation can be drawn from Table 4.4. The number of CS into Scots is naturally higher in the volumes where more Latin is used, i.e. volumes one to six. In volumes seven and eight, on the other hand, CS into Latin outnumber those into Scots, which is not particularly surprising given that more than half the entries are in Scots in these two volumes. What is striking is the relatively low number of CS in volume two (62.4 CS per 1,000 entries) and volume six (67.2 CS per 1,000 entries). The three highest relative numbers of CS can be found in volumes four (90.8), 5.1 (118.8) and 5.2 (127.4), i.e. at the beginning of the increase of Scots. While these figures neither take the length or the nature of the CS nor entries with �multiple languages’ into account, they suggest that the number of CS is higher at the beginning of the vernacularisation process. However, statistical tests indicate that there is no statistically significant correlation between the decrease of Latin entries and the number of CS in the first eight volumes of the Aberdeen Council Registers.41 The number of entries with Latin as the matrix language decreases steadily, while there is a spike in CS in volumes 4–5.2. The spike in CS into Scots may be due to scribes starting to consider Scots as a language that was appropriate to use in legal contexts or even necessary to aid comprehension (see below). CS into Latin, on the other hand, may have been used in Scots entries to stick to the traditional structure and conventions of the registers, which may have been considered more important in the earlier volumes than when Scots was fairly well established as a written legal language. Without a careful qualitative analysis of the nature of the CS in all volumes of the council registers, these possibilities remain speculations. The following analysis of CS in volume 5.2, in which the highest relative number of CS can be observed, does, however, support some of these suggestions.
Table 4.4 shows that 46 CS are into Latin and 88 CS into Scots in volume 5.2. These CS can be divided into (a) intrasentential CS, i.e. switches between or within sentence constituents, and (b) intersentential CS, i.e. switches between sentences or independent clauses.42 The majority of CS in volume 5.2 are intrasentential (34 into Latin and 87 into Scots). Intersentential code-switching is more common in switches into Latin because Latin is sometimes used for the heading and/or date at the beginning of entries. Entry ARO-5-0782-01, for example, starts with the date in Latin before continuing in Scots, as the following transcription and image show (Figure 4.4):
FIGURE 4.4 Entry ARO-5-0782-01 of the Aberdeen Council Registers, dated 11 November 1454. By permission of Aberdeen City and Aberdeenshire Archives.
The image illustrates that there is no difference in the scribe’s handwriting between the Latin beginning and the Scots continuation of this entry. In fact, no change in handwriting was detected for any of the CS in volume 5.2. There are, however, other visual cues in entries that switch between Scots and Latin, such as line breaks. Kopaczyk’s model for conceptualising code-switching on different levels takes such visual cues into account and highlights the interaction between these cues and instances of code-switching.43 In this model, CS can be categorised on six different levels: (a) macrogenre: the switch of codes between different genres (i.e. what Voigts refers to as intertextual level),44 (b) discourse: CS beyond the sentence level (i.e. for �larger, functionally and textually coherent passages’ such as reported speech), (c) clause: CS between complete clausal units (i.e. intersentential code-switching), (d) word: CS for a single phrase or word (i.e. intrasentential code-switching), (e) morpheme: the use of a different code for single morphemes (e.g. for the root and for inflections), and (f) orthography: when words are spelled �according to orthographic conventions taken from another code’.45 The model then allows researchers to specify whether the scribe used visual cues for these CS. Applied to volume 5.2, CS in the Aberdeen Council Registers can be described in the following way based on Kopaczyk’s model. Since the entries included in the council registers can be considered to constitute one text genre (legal texts), there are no CS on the macrogenre level. Code-switching can, however, be observed on the discourse, clause, word and morpheme levels.46 There are no visual cues for CS on the word and morpheme level. On the discourse and clause level, on the other hand, the scribes do use visual cues in the form of line breaks. This is, however, not always the case, as the example of the Latin date earlier shows.
Dates constitute only one part of the CS into Latin, as Figure 4.5 illustrates. Most CS into Latin are modifications of names, such as pater (father) or filius (son). After dates, the next most frequent use of code-switching into Latin are Latinised names, such as Willelmus for ’William’, which can be considered CS on the morpheme level. There is also some legal terminology for which Latin is used, e.g. �transeat de villa’ for �[s/he may/must] pass through the town’. As mentioned before, Latin is also used in headings and marginalia. The copy of a letter in Middle Dutch is, for example, introduced in Latin: �Copia obligacionis Walteri michaelis de Andwarpia’ (ARO-5-0714-02). Similarly, Latin is used as a structuring element and as a cohesion marker, with expressions such as �in primis’ or �videlicet’.
FIGURE 4.5 Code-Switches (CS) into Latin in volume 5.2 (1441–1472).
In summary, CS into Latin are mainly used for specific legal writing practices, such as modifiers after names, and to structure the text through dates, headings and other structural elements. CS into Scots, on the other hand, seem to be used when there is no Latin equivalent, when the Latin equivalent is unknown or much rarer than the Scots term, or to ensure understanding. Figure 4.6, which shows the CS into Scots divided into different functions, illustrates these points.
FIGURE 4.6 Code-Switches (CS) into Scots in volume 5.2 (1441–1472).
In contrast to the CS into Latin, most CS into Scots refer to locations or geographical features, particularly place names, names of streets, crofts and rivers. To provide just one of the 40 examples, entry ARO-5-0699-01 reads �Jacentem ex parte occidentali de ly schipraw inter terram Johannis Rede […]’, with �ly schipraw’ referring to a street name in Aberdeen that is still in use today, albeit in its English form �shiprow’. This is a typical example of intrasentential CS. The Scots form of this particularly street name was, however, not always used. In fact, the Latin equivalent (�Vico Nauium’) occurs more frequently than �schipraw’, but only in entries with Latin as the matrix language (in 179 entries). Overall, the Scots form is used in 21 Latin and 41 Scots entries, and is particularly frequent in volumes six and seven of the council registers. However, no particular development in the use of the Latin as opposed to Scots form emerges from this data. Geographical terms, like street names, and Scots surnames together account for about three quarters of all CS into Scots. While first names are often Latinised, surnames usually remain in their Scots form, such as �alexander of kintor’ in entry ARO-5-0652-04. Furthermore, some objects (for example, �ly barge’) as well as some measurements and currency (such as �mitez’, i.e. a small copper coin) appear in Scots.
As these examples show, the Scots terms are often introduced with the definite article �ly’, which was borrowed from Anglo-Norman. The Dictionary of the Scots Language explains that �ly’ (also spelled le, lie or lee) was used �when citing, in Latin contexts, appellative place-names in the vernacular which idiomatically required the definite article’, and possibly subsequently was �applied also to any vernacular designation in a Latin context’ and was similarly used in medieval English materials.47 The use of the Anglo-Norman article shows that Scots was influenced by a variety of languages, many of which were used in a legal context, demonstrating the multilingual nature of this text type.
A few CS into Scots can also be found for accounting and legal terms (e.g. �ly tollis’, which refers to tax paid to a landowner), occupations (such as �baxter’ for �baker’) and actions (e.g. �Johannes Richardson grauntis plukkyng of schorlingis’ for �John Richardson grants the gathering of sheep skin’). Other CS into Scots were difficult to categorise (see �other’ category). Sometimes an entry starts in Latin but ends in Scots, as the following transcription of entry ARO-5-0676-01, which deals with fish trade, illustrates. This entry was annotated with the attribute “mul” for multiple languages, as it is difficult to determine the dominant language of this entry. Therefore, no �foreign’ annotations are used here, even though the scribe clearly switches between Latin and Scots. Where exactly these CS start and end is, however, not always easy to determine. The abbreviation barrell’ could, for example, be interpreted as Latin or Scots. Indeed, it remains unknown if the scribe himself consciously distinguished between these two languages (see Trotter’s earlier statement). There is certainly no indication in the handwriting that suggests that the two languages were strictly divided (see Figure 4.7). This kind of mixing of languages is not unusual in late medieval Britain and can be observed in various text types. Wright, for example, illustrates that the abbreviations in late medieval business writing in England can simultaneously be interpreted as both English and Latin.48
FIGURE 4.7 Entry ARO-5-0676-01 of the Aberdeen Council Registers, dated 9 March 1444. By permission of Aberdeen City and Aberdeenshire Archives.
Particularly interesting in the previous entry is that the scribe translates a Latin phrase into Scots: Et debit’ p[ro] pena earu[n]d[em] q[uod] no[n] veneru[n]t p[ri]mo a[n]no to And for the payne that thai come noght the first yher’. This shows that there was a Latin equivalent to the Scots phrase and that the scribe was happy and ready to use it. It can, therefore, be assumed that this translation is meant to aid comprehension for the readers of the records.
The previous entry demonstrates the multilingual nature of the council registers. Of course, these records are no exception to the use of various languages in other texts. Amongst others, Wright, Schendl and Kopaczyk have illustrated that multilingualism was the norm in late medieval and early modern texts.49 The potential correlation between vernacularisation and CS, however, remains under-researched. The quantitative analyses presented here suggest that CS is more common at the beginning of the vernacularisation process. While the data from the Aberdeen Council Registers do not provide evidence for a statistically significant correlation, the results still indicate that the scribes switched between languages more frequently when they started to use Scots as the matrix language in entries more often. To fully understand why this is the case, instances of CS in all eight volumes of the registers would have to be analysed qualitatively, which needs to remain a topic for further research at this stage. There are, however, other conclusions that can be drawn from the results presented here.