Multilingual practices in women’s English correspondence 1400–1800

Submitted pre-print. Published in Sebba, Mark et al. (eds.) 2011. Language Mixing and Code-Switching in Writing. Routledge.

Multilingual practices in women’s English correspondence 1400–1800 ARJA NURMI AND PÄIVI PAHTA

1. INTRODUCTION This paper explores multilingual practices in Englishwomen’s personal correspondence in 1400–1800.1 In the course of the 400-year period in focus in our study, the sociolinguistic situation in England underwent considerable changes, well known from general histories of the country. At the beginning of the period, England can be characterized as a diglossic, or triglossic, society where several languages were used with partly overlapping social functions. Latin and French were used in institutional contexts and among the upper social strata, while English was a lowprestige local language, primarily used in spoken interaction (for discussion, see e.g. Nurmi and Pahta, 2004; Pahta, 2004, Schendl and Wright, forthcoming). During the fifteenth century, members of a French-speaking Norman aristocracy were still present in the society, while at the same time, English gradually began to emerge from the shadow of French and Latin. At the end of our focus period, the language situation was very different. English was a fully-fledged national language, also beginning to obtain prestige functions as an international language, e.g. in scientific writing (see e.g. Pahta, 2011). This complex societal multilingualism and its diachronic change is not, however, in focus, nor really in evidence, in our study. Rather, the focus of our attention is the ways in which our female English informants of the period draw on their linguistic resources in classical and contemporary European languages, learned through teaching instead of by living in a multilingual community. The study is corpus-linguistic, combining quantitative and qualitative methods of analysis. Our data comes from the Corpora of Early English Correspondence (CEEC400), containing over 5 million words of running text sampled from letters written between 1402 and 1800. At the outset, the data we draw on can be characterized as monolingual English, as the corpus was specifically compiled of texts representative of the English language to facilitate sociolinguistic research of its history. As we have shown in earlier research, despite this allegedly monolingual nature, the letter corpus, like many other historical English corpora, provides ample evidence of code-switching between English and other languages. While our earlier studies examine code-switching in the letter corpus either in a broad general perspective (Pahta and Nurmi, 2007), according to the writers’ social rank (Nurmi and Pahta, 2004), or in the correspondence of individual male writers in interaction with their social networks or in their varying social roles (Pahta and Nurmi, 2009; Nurmi and Pahta, 2010), our present concern is women’s code-switching practices. We aim to provide an overview of ways in which women writers of the period in focus make use of their multilingual resources in their personal correspondence, and compare our findings with those observed in men’s contemporary writing practices. Furthermore, we make comparisons with results based on other types of data to contribute to the general theory-formation on code-switching in writing.

The structure of the article is as follows. Section 2 discusses women’s literacy and education in the period of the study, i.e. the historical preliminaries providing access to multilingual resources for women. Section 3 presents our material and method in more detail. Section 4 explores multilingual practices evident in the language of our informants, focusing particularly on textual and discourse functions. Section 5 presents our conclusions. 2. LITERACY AND EDUCATION Women’s literacy, as well as their educational opportunities, changed along the four centuries covered in this study. This is clearly evident from the nature of the correspondence data available in CEEC400 that aims to represent the literate social strata of the period it covers as accurately as possible: during the fifteenth century many of the letters written in the name of a female writer were in fact put down to paper by male scribes, while by the seventeenth century almost all women in the corpus were in the position of writing their own letters (Nevalainen and RaumolinBrunberg, 2003; Nurmi, forthcoming). The inequality in educational opportunities existed not only between different social strata, but also between the genders. Higher education was only available to women in private, as universities were a solely male domain. Despite this, there were individual women who, through their social status, like Queen Elizabeth, or through the support of a learned father or other mentor, like Thomas More’s daughter Margaret Roper, received an education that in addition to the principles of Christianity and basic literacy and numeracy also comprized more advanced learning, including classical and foreign languages. During the eighteenth century, as the ideals of the Enlightenment made learning something to be sought after, there were more books for self-directed studies, so that women like novelist Fanny Burney or governess Agnes Porter could also teach themselves languages and educate themselves in other topics (for Fanny Burney, see e.g. Nevala and PalanderCollin, 2010; Pahta and Nurmi, 2009; for Porter see Nurmi and Nevala, 2010). Indeed, it can be argued that it was the more widely spread literacy during the eighteenth century that allowed women to become multiliterate in not just their native English, but also in other languages they deemed interesting or useful. At the beginning of the period we are studying, the circumstances were entirely different from those at the latter end. It is difficult to estimate the rate of women’s literacy in the fifteenth century. Some indication of women’s ability to read can be gained from their wills, which often bequeathed devotional books to female relatives (Keen 1990, 220). Autograph wills have been used as evidence of women’s writing ability, but this probably gives an underestimated degree of literacy for women, since women were less likely to leave wills than men (Laurence 1994, 165). At this stage, as later, women’s literacy depended on their social status, but also on their place of habitation. Londoners were much more likely to be literate than their rural counterparts (Laurence 1994, 166). Between 1580 and 1729, for example, women’s literacy in London increased from 16 per cent to 56 per cent, while their counterparts in East Anglia began from zero and reached 26 per cent literacy in the early eighteenth century (Cressy 1980, 144). When looking at the average for Englishwomen, around 1500 approximately 2 per cent were literate, almost reaching 10 per cent by 1600, remaining below 30 per cent in 1700 and even in 1800 estimated at just over 40 per cent (Cressy 1980, 177; for a discussion of problems in these estimates see Brink 2010, 28–29).

Education in the fifteenth century was mostly only available to the higher ranks of society. For girls, the opportunities were significantly narrower than for boys. There were some similarities, as, for example, it was still common to send both boys and girls to be educated in the households of great lords and ladies. An option that apparently was more common for girls than boys was education at religious institutions. In fact, it seems that sending daughters to a convent for their education was as frequent as, if not more common than, education in a great household. Evidence for this practice comes from the means some bishops took to limit the number of girls to be educated at convents, since it was felt they were keeping the nuns from their devotional duties. The teaching girls received would not offer any great depth of learning, but interestingly for our study they would learn to read and be instructed in the elements of French. During the late Middle Ages schools were beginning to multiply, as benefactors sought to finance them through wills and other donations. At least some of the stipulations mentioned ‘children of the parish’, so it is possible this included girls as well as boys. Overall, elementary schooling became increasingly better organized and more available to a wider proportion of the population (Keen 1990, 228–30). There were better and worse times, however, since, for example, after the dissolution of the monasteries (and convents), grammar schools were established to provide education for boys, but no similar provision, with some exceptions, was generally made for girls (Laurence 1994, 167–8). For much of the time under study, women’s education was viewed as useless, since their main role was perceived to be as wives and mothers. Many people also felt that educating women could be dangerous. Richard Hyrde discusses these attitudes in his introduction to Margaret More’s (later Roper) translation of a treatise by Erasmus in 1527, mentioning that women knowing Latin and Greek was ‘not only neither necessary nor profitable, but also very noisome and jeopardous’ (cited from Cressy 1975, 107). It was felt that being able to read texts written in Latin and Greek would ‘inflame their stomachs a great deal the more to that vice that men say they be too much given unto of their own nature already’, that is, women would forget their place and get ideas better befitting men. As the Tudor reign went on with queens as heads of state, it apparently became more prudent to soften expressed attitudes towards the education of women. In 1581, Richard Mulcaster wrote that ‘naturally the male is more worthy and politically he is more employed, and therefore that side claimeth this learned education as first framed for their use and most properly belonging to their kind’ (cited from Cressy 1975, 110). He went on to say that women should receive some education, according to their degree. ‘We see young maidens be taught to read and write and can do both with praise; we hear them sing and play and both passing well; we know that they learn the best and finest of our learned languages to the admiration of all men’, discussing the benefits of learning different skills and admitting that ‘[r]eading if for nothing else … is very needful for religion, to read that which they must know and ought to perform’; the need for foreign or classical languages is not something he discusses. Nearly a hundred years later, education for women was still a contentious issue. John Amos Comenius was pro-education: ‘Nor can any sufficient reason be given why the weaker sex (to give a word of advice on this point in particular) should be altogether excluded from the pursuit of knowledge, whether in Latin or in their mother tongue’ (cited from Cressy 1975, 111). However, he went on to qualify this opinion, stating that ‘[w]e are not advising that women be educated in such a way that their tendency to curiosity shall be developed, but so that their sincerity and contentedness may be increased, and this chiefly in those things which it becomes a

woman to know and to do; that is to say, all that enables her to look after her husband and to promote the welfare of her husband and her family’. There were schools providing women with a broader education, as the example of Bathusa Makin shows. In 1673, she started an academy for young ladies, and the selection of languages taught included Latin, French, Greek, Hebrew, Italian and Spanish (Cressy 1975, 112). In the dedication to a volume on women’s education Makin writes: ‘I verily think women were formerly educated in the knowledge of arts and tongues, and by their education many did rise to a great height in learning. Were women thus educated now I am confident the advantage would be very great. The women would have honour and pleasure, their relations profit, and the whole nation advantage’ (cited from Cressy 1975, 113). Teaching was also becoming a possible career for women from the mid-seventeenth century onwards, in the private schools for the daughters of gentry and later on as governesses (Laurence 1994, 171). By the eighteenth century, education was more freely available than at the beginning of our period of study, but still girls (and the rural population) were at a disadvantage. Primary education continued to focus on reading, writing and arithmetic, and classical teaching was offered at grammar schools, which were not accessible for girls. More typically, there was an emphasis on employable skills for the lower strata of society (spinning, weaving and knitting for girls), while morality was a constant (Black 2001, 96–7). Girls were more commonly educated at home than boys, although schools for girls became more frequent during the eighteenth century (Porter 1991, 25, 164). For girls in the highest social echelons much emphasis was placed on manners and deportment, as well as accomplishments of all sorts, rather than formal learning. Nevertheless, also French, history and geography, to name a few subjects, would be taught to young ladies of quality (Porter 1991, 165; Martin 1998, 52). 3. MATERIAL AND METHOD: CORPUS-BASED ANALYSIS OF PERSONAL CORRESPONDENCE The material for this study comes from the Corpora of Early English Correspondence (CEEC400), altogether 5.2 million words of personal correspondence written between 1402 and 1800 (approximately 12,000 letters by over a thousand letter writers). The corpus used here combines three parts: the original Corpus of Early English Correspondence (CEEC), its supplement (CEECSU) and extension (CEECE). All three have been conflated into one large corpus for our study. (For more information on the corpus, see Palander-Collin et al. 2009: 14–16 or the CoRD entry; for a list of letter collections included see the Appendix in Nurmi et al., 2009.) Because of women’s low literacy rates compared to men, and because of editorial principles prioritizing male writing, only approximately 20 per cent of the whole corpus consists of women’s writing. All in all, the data analysed in the study contains letters by 292 female writers, making up over one million words of running text. Table 1 gives the details by century. Not only is the amount of women’s writing clearly increasing over time, but there is also a marked rise in the number and proportion of women whose letters contain code-switching in the eighteenth century. Our working definition of code-switching comes from Heller (1988, 1): ‘the use of more than one language in the course of a single communicative episode’. As our study covers a long timeperiod, the questions of borrowing vs code-switching are even more thorny than in a strictly synchronous study. We have taken an inclusive

approach, preferring to track all elements with a trace of foreignness to them, rather than ruling out potential switches. Table 1. Corpus of Early English Correspondence (CEEC400) and number of women code-switching. Time words female writers code-switchers % code-switchers 1402–1499 65,391 33 5 15.2 1500–1599 76,231 61 5 8.2 1600–1699 362,618 112 14 12.5 1700–1800 586,138 86 25 29.1 Total 1,090,378 292 49 16.8 In this study, the instances of code-switching in the data were located using text-level annotation for foreign language which was coded into the corpus during compilation and is illustrated in (1).2 (1)

I see every Body but converse with nobody but (\des Amis choisies\). In the first Rank of these are Lady Stafford and dear Molly Skerret (Montagu: 1725?, Mary Wortley Montagu, II,44)

The bracket annotation was used as the search parameter for retrieving the codeswitching data, which ensured efficient and systematic data extraction, characteristic of the corpus analysis of even large datapools.3 An alternative method would have been to use an alphabetical wordlist of the corpus texts for identifying foreignlanguage items in the data; this method, less automated and more time-consuming, can also be used in analyzing code-switching in historical corpora that do not contain text-level annotation for foreign language (see e.g. Pahta, 2004; 2007; 2011). In addition to the passages coded in the corpus as switches, some longer foreign language passages that have been omitted from the texts during the compilation in CEEC400 were retrieved from the source texts and included in the analysis. These passages too were located using the code indicating omission (e.g. [^FRENCH OMITTED^]) as the search parameter.4 Since we are more concerned in this study with the writers’ textual and discursive uses of their multilingual resources than with the grammar of switching, in the quantitative analysis we have counted code-switched segments, i.e. passages in a language other than English, in the data instead of individual switches from one language to another. Thus a French passage in a text is counted as one switch, even though there are in fact two switches, first from English to French and then from French back to English. In an attempt to measure the frequency of foreign-language passages, we have counted the number of switched passages and normalized the frequency /1,000 words to facilitate comparison across centuries and with other datasets. This practice is in accordance with our earlier studies. Table 2 presents the results of these calculations in women’s letters in CEEC400. In addition to the frequency of code-switched segments, we have also studied the length of the switched passages in orthographical words, i.e. words separated by spaces. This means that, for example, compounds written with a space in between their two parts are counted as two words. This analysis provides us with another indicator of foreign-language density in the data, average switch length (ASL in words), that can also be used in the comparative analysis; this is also a technique we have employed in previous research.

Table 2. Number of code-switched passages (CS) in women’s letters in CEEC400, with normalized frequencies /1,000 words and average switch length (in words). Time words CS CS/1,000 w. ASL 1402–1499 65,391 10 0.15 2.9 1500–1599 76,231 28 0.37 4.9 1600–1699 362,618 63 0.17 3.5 1700–1800 586,138 564 0.96 4.1 Total 1,090,378 665 0.61 4.0 The results of table 2 give us a clear indication of the increasing frequency of codeswitching related to the number of words in the eighteenth century as opposed to the previous centuries. The number of switched passages increases vastly, and this tendency is clearly seen when the number of switched passages is related to the number of words in each century. The slight increase in the sixteenth century is due to Queen Elizabeth’s letters abroad, which have the address outside in French (see section 4.2.1). Queen Elizabeth is in fact responsible for the vast majority of switches in the sixteenth century, 21 of the total 28, and 9 of the instances (32.1 per cent) are addresses. Overall, these frequencies are clearly lower than those attested in men’s letters: 0.71 /1,000 words for men in general in 1410–1550 (Nurmi and Pahta 2004, 435), 3.0 for Charles Burney and 4.9 /1,000 words for Thomas Twining in the eighteenth century (Pahta and Nurmi 2009, 37). The average length of switched passages does not seem to vary greatly, although there is some increase as time goes by. Since addresses written on the outside of letters typically consist of several words, it is not surprising that Queen Elizabeth’s letters abroad influence also this metric in the sixteenth century. There are 135 foreign-language addresses in our total data (20.3 per cent), most of them in the eighteenth century when switching is at its most frequent. Because of this, the influence of addresses on the average switch length is at its most notable in the sixteenth century, when switching is rare. 4. MULTILINGUAL PRACTICES IN WOMEN’S CORRESPONDENCE The intersection of multilingualism and gender reflects the particular society in which multilingual interactions take place. (For a good summary of earlier research, see e.g. Mills, 2006). While our main concerns for women’s multilingualism during the four centuries studied have been literacy and availability of educational opportunity, it is interesting to note that these have been relevant concerns also in studies of presentday societies (see e.g. Sadiqi, 2008 for the usage of Modern Standard Arabic, French, Berber and Moroccan colloquial Arabic in Morocco). As discussed in section 2 above, the availability of education was one of the factors which limited women’s linguistic repertoires in our data, as their opportunity of becoming fluently multilingual in any way was much more constrained than men’s. In present-day communities the relevant questions may be slightly different, ranging from the mastery of specific registers and achieving a high degree of biliteracy in more than one language (see e.g. the case of Latinas and academic Spanish in Guerrero, 2003) to the maintaining or shifting of linguistic identity as immigrants respond to the requirements of their social networks and other influences (Stoessel, 2002). The use of switched passages as expressions of identity, typically membership of a particular group (in our data usually the educated elite or the aristocracy), is also a trend attested in present-day speech communities. While most present-day work focuses on societal multilingualism, and the active maintenance of bilingualism (on the resistance to monocultural and monolingual ideas

of a bicultural and bilingual writer, see Kraver, 1997), there is also some work done in the field of second-language learning (Pavlenko et al., 2001). More complex identitybuilding, such as the humorous narratives of bilingual women, creating links through incomplete L2 skills (Bell, 2007), could not be found in our data, although there are instances where something resembling this could be interpreted as present (see section 4.2.1). There was also no room in this study to see whether the trend Valdés-Fallis (1978) observed among bilingual Mexican-American women to associate more closely to the code-switching style of male interlocutors than female ones would hold true in this data. The multilingual practices in women’s correspondence can be regarded from many points of view. As our previous research suggests that the most valid details in describing differences between individuals in general can be found in the frequency of code-switching, measured here as the number of switched passages per 1,000 words, and in the selection of languages used, those are two points on which we focus. From a more functional point of view, the difference between primarily using preformulated chunks, whether quotations, conventionalized expressions or terminology (which could be regarded as borrowings), and more creative use of multilingual resources, which testifies to a higher level of fluency in the language employed, is the most salient one. We begin by discussing the languages besides English found in use by women in our data, and continue to the functions of switching after that. These two issues are somewhat related, as some types of switching would most naturally appear in a specific language. 4.1. Languages in use Besides English, there are instances of several other languages in the correspondence of our informants. As could be expected, the selection of languages in use is narrower than that in contemporary male writers’ letters (cf Nurmi and Pahta, 2004, 2010; Pahta and Nurmi, 2007, 2009). Women drawing on multilingual resources in their letters are typically gentlewomen or members of the aristocracy, with only few exceptions. The three languages appearing most frequently in women’s letters are French, Italian and Latin. French shows a most obvious rising trend across the centuries, and is generally the best known language; thus it also tends to be the resource that our writers draw on most creatively (see below, particularly section 4.2.6). The appearance of Latin fluctuates somewhat, and does not reach a notable peak at any point. This is probably because classical education for women was so rare, and highly educated women needed not only the required linguistic skills but also the appropriate correspondents with whom to include Latin passages in their letters (cf Pahta and Nurmi, 2009 and Nurmi and Pahta, 2010 on the reciprocity and audience design of code-switching). Table 2. Languages in use in women’s letters in CEEC400. /1000 w Italian /1000 w Latin Time French 0.03 0.00 1402–1499 2 0 8 0.17 0.00 1500–1599 13 0 14 0.11 0.00 1600–1699 39 0 22 0.79 0.07 1700–1800 462 41 56 0.47 0.04 Total 516 41 100

/1000 w 0.12 0.18 0.06 0.10 0.09

Total 10 27 61 559 657

Italian is only found in the eighteenth century, and apart from isolated instances of highly conventionalized expressions appears mostly in the language of three eighteenth-century women: Lady Mary Wortley Montagu, Hester Lynch Piozzi and Fanny Burney. There are clearly identifiable language-external reasons for this in the circumstances and life-style of all three. Montagu lived in Italy for a long time, yet it is only a few phrases of Italian that appear in her letters, suggesting the lack of fluency in the language by her correspondents or possibly Montagu herself. Hester Lynch Piozzi’s second husband, Gabriel Mario Piozzi was Italian, so there was a very personal reason for the appearance of the language in her letters, and her correspondents obviously understood enough of the fairly simple phrases. Fanny Burney was familiar with Italian partly through her work as an amanuensis to her father, Charles Burney; he knew Italian for his work on the history of music (see Pahta and Nurmi, 2009). Other women using Italian passages were all equally members (or hang-around members) of the literary elite. In addition to the three languages mentioned in the table, there were isolated appearances of German, Malay and Greek (examples 2–4). In (2) Princess Charlotte, daughter of George III is discussing events in Germany, and applying a German term when referring to the central cities of administrative units. Eliza Draper describes her life in India (3), where she has been living for two years at the time of writing this letter. The cultural and political conflicts are in evidence in her letters, and the practice of running amok is not something she has witnessed first hand, but is rather a cautionary tale intended to prevent her from riding alone. (2)

The country people are uncommonly good and behave very well, which is very much to their credit when it is known how cruelly the French have behaved, destroying the 13 best Oberaemter of the Dutchy and taking away money, corn, cattle; (George3: 1799, Charlotte Hanover, IV,47)

(3)

They have a Custom too of running a Muck at a particular season of the year, which is being intoxicated with opium, & then making a merit of slaying every Christian they meet (Draper: 1769, Eliza Draper, 99)

Example (4) bears witness to the Greek skills of Elizabeth Carter, who was known for her literary and scholarly qualifications. For example, Samuel Johnson reputedly praised a Greek scholar, saying that he spoke Greek better than anyone he had met, excepting Elizabeth Carter (Hawley, 2009). (4)

Some indeed were so complaisant as to affirm that I could not have performed so arduous a Task [part deleted] as ascending the Hill without the Assistance of Pegasus, but this polite Speech was soon contradicted by others who gave me the frightful Greek Name of Aerobates which they maliciously translated, riding [part deleted] like a witch upon a Broomstick. (Carter: 1739, Elizabeth Carter, 69)

However, the example also shows that the life of a female scholar was not easy in the eighteenth century, and that there were people prepared to mock her accomplishments (even if in jest here), using Greek terms to do so and, in a manner of speaking, turning her own knowledge against her. Aerobates (άεροβάτης) was a term found in Hellenistic Greek, meaning ‘person who walks the air’ (OED s.v. aerobat). Carter’s

critics turned this into a misogynistic reference to witchcraft, which is a traditional way of denigrating learned women. 4.2. Functions of code-switching Multilingual resources are mobilized for a variety of functions in our data. The foreign-language passages can be purely practically motivated, reference items or concepts from other cultures, or they can create intertextuality or express identity. Knowledge of foreign languages was linked to education, and could be displayed as an indication of membership in the elite. Governess Agnes Porter reports in one of her letters an interesting discussion concerning the education of women. Her travel companions were a glover, hatter and grocer, as well as a ‘Miss from Sherbourne school’, who set out to display her accomplishments. (5)

The young woman immediately began one – sang it in a very particular style, and when it was over said to me: ‘Shinty vue, Madame?’ I told her that I should understand her better if she spoke English. ‘La, Ma’am, how I pity you! What, not speak French? I would not give up that accomplishment for the world – well, it certainly is your misfortune, Ma’am.’ ‘As to misfortune’ said the grocer ‘I do not see much in it – this gentlewomen here seems a quiet, steady person, and most likely is kind and obliging to her husband and careful to keep her children, if she has any, right and light, which I think if she does so is better than to parly franchee as you call it, or such like. As for my part, I think a woman can talk quite enough in her mother tongue, and has no need to dun folks with a noise in more languages than one, as they have words enough and to spare in plain English.’ The hatter and Miss were shocked at such assertions; the glover coincided with the old man, and their arguments I thought amusing on each side of the parties. When the young woman and I were by ourselves I told her in French that, as I supposed our fellow travellers did not understand that language, I thought it better to decline the pleasure of answering her in French. She looked a little serious, and replied in English ‘Ma’am, I am not so far in the French phrases’! (Porter: 1789 Agnes Porter 81)

As this anecdote illustrates, in the eighteenth century knowledge of French was a desirable attribute for people with aspirations. Since it was addressed to one of Porter’s pupils, Lady Mary Fox Strangways, it further illustrates the mores she is trying to instill in her wards, and, perhaps more interestingly, also allows Porter an opportunity of affirming her own social position. As opposed to the upstart ‘Miss from Sherbourne school’, Porter is asserting her own status as firmly in the camp of her aristocratic ward: there are the people who know proper French and also the proper time for speaking it, and then there are the hoi polloi, who have ideas above their station. It should perhaps be noted here that Porter was indeed quite fluent in French herself, and on occasion wrote longish passages of her letters to her pupils in French as a means of instruction. (For more on Agnes Porter’s life and language, see Nurmi and Nevala, 2010.) 4.2.1. Genre-specific conventions

As with our earlier studies concerning code-switching patterns in correspondence, women’s letters contain frequent genre-specific conventions where switching occurs. The most frequent of these is the address outside the letter, but also dates, as well as opening and closing formulae have been attested (see Nurmi and Pahta, 2004; Pahta and Nurmi, 2007). Almost one fifth of all foreign-language passages in our present data are addresses, either outside the envelope as in (6)–(8) or instructions for writing an address inside the letter as in (9) The addresses can be either fully in another language, or they can be partially English, partially French. (6) is the only fifteenthcentury example; it is also the only one which can be regarded as a possible example of Anglo Norman. All the other instances are in letters sent to or from abroad. (6)

A moun trescher et bien amé cosin William Paston soit donné. (Paston: c. 1426, Elizabeth Rothenhale, II,2)

(7)

A monsieur mon bon frere et cousin le roy d’Escosse. (Royal1: 1585, Elizabeth I, 12)

(8)

A Monsr. Monsr. de Preaumont. (Basire: 1651, Frances Basire, 107)

(9)

Be pleas’d to direct recommandé a Signor Belloni, Banquier a Rome. (Montagu: 1741, Mary Wortley Montagu, II,224)

Queen Elizabeth writes to James the VI of Scotland in English, but the address outside the letter (as in 7) is frequently in French. This may have been a diplomatic convention of the time, as it is hard to imagine any messenger between the two courts would have been unable to read English or Scots. Even poorly educated women, such as Frances Basire, a clergyman’s wife (8), were able to address their letters in French. It is quite plausible that her husband has sent her instructions on how to address the letters, and Basire herself did not need any skill in French to follow such instructions. In fact, it was quite common to instruct people on how to address letters sent to the continent (9), whether because of changing addresses or presumed lack of language skills. French seems to have been the conventional language for addresses at all times regardless of where the letter was sent to or from. It should be noted, however, that our data does not always show addresses. So, for example, Eliza Draper’s letters from India have no address included in the available edition, and we have no way of knowing whether the language used was English or something else. In fact, all the addresses in women’s letters show correspondence within Europe. Mary Wortley Montagu, writing from anywhere on the continent would address her letters either in French or a mixture of French and English (10), and Hester Lynch Piozzi mixed French and English when writing from Italy (11). A rare example of English people writing to each other in France is in Mary Wollstonecraft’s letter written in Paris and addressed to a friend (carefully referred to as Citoyenne Barlow) also in Paris (12). (10) To The Rt Honble The Countess of Bute recommended to Fran: Child Esqr near Temple Bar London Angleterre par Amsterdam (Montagu: 1758, Mary Wortley Montagu, III,142)

(11) A Monsieur / Monsieur Samuel Lysons / chez le Reverend / Mr. S. Peach / a East Sheene / near Mortlake / Surréy. / Angleterre. (Piozzi: 1785, Hester Lynch Piozzi, I,167) (12) A la Citoyenne Barlow. / Hotel de la Grande Bretagne. / Rue Jacob. / Faubourg St Germain. (Wollstonecraft: 1793, Mary Wollstonecraft, 234) Another genre-specific convention found in the letters of men, writing dates in another language (Nurmi and Pahta, 2004), is quite rare in women’s correspondence. There are altogether six instances of this, most of them references to holidays (13). Four of the dates are in Latin, and only two in French (14–15). The latter of the French examples illustrates nicely the importance of the recipient for linguistic choices in letters: in (15) Elizabeth Twining is writing to her young nephew, who has been studying French with her. This way of dating the letter in French reminds young Daniel of their shared hours of study, and is perhaps also an example of a humorous way of maintaining the bond between them. This is also the only date in a language other than English in the eighteenth-century data. In fact, four of the instances are found in the fifteenth and early sixteenth centuries, and there is a lonely Anno Domini in the seventeenth. (13) Wretyn at Paston in hast þe Wednesday next after Deus qui errantibus,5 for defaute of a good secretarye, &c. Yowres, Agnes Paston (Paston: 1440?, Agnes Paston, I,26) (14) Jesu spede you ever, amen. At London le xviij jour d’aoust. Elysabeth Stonor. (Stonor, 1476, Elizabeth Stonor, II,9) (15) Fordham, le 24ième de Décembre, 1764 (Twining, 1764, Elizabeth Twining, I,55) Other genre-specific conventions include letter openings and leavetakings, but these are not particularly frequent either (eight altogether, in the seventeenth and eighteenth centuries), including phrases such as Mon cher Ami and Adeiu, je suis á vous de tout mon coeur. 4.2.2. Quotations, proverbs and fixed phrases Typical instances of switched passages include quotations, proverbs and other preformulated fixed phrases. Quotations may be of either well known authors or of familiar persons, although the latter is far less frequent. There are altogether 55 passages that we have tentatively identified as quotations, although the difference between a quotation and proverb is difficult to establish; 32 of the quotations are in French, 19 in Latin and four in Italian. The French quotations are often either somewhat euphemistic expressions, as Carter’s reference to the effects of migraine in (16), or they are references to the French-speaking culture of the upper echelons of society (17), although here too the euphemistic interpretation is possible, since discussion of strong emotion may not have been the done thing. (16) & I agree with Carraccioli who says on est bien sot, lorsqu’on a la migraine. (Carter, 1791, Elizabeth Carter, 173)

(17) There is a fine lady there too, whom I hear the French say “qu’il aime avec passion,” & he don’t deny it, (Lennox, 1777, Sarah Lennox, I,257) The Latin quotations are typically from classical sources, and do not differ in form and usage. So Margaret Roper’s Juvenal quote in (18) works quite similarly to Arabella Stuart’s Virgil quote (19). In Roper’s case, the use of Latin in the letter to her father (incarcerated in the Tower) can also be interpreted as a means of reminding him of happier days when he was teaching her the language. Mary Wortley Montagu’s passage is a quotation from Abraham Cowley’s Epitaphium Vivi Auctoiris (Epitaph of the Living Author) (20).6 Her switched passage is perhaps more integrated in the actual content of the letter than in the previous two, since Montagu is interpreting the word pauperie in reference to her own situation and inclination in life. (18) as becommeth a very true worshipper and a faithful seruaunt of God, which I doubt not, good father, holdeth his holy hand ouer you and shall (as he hath) preserue you both body and soule (vt sit mens sana in corpore sano) and namely, now when you haue abiected all erthly consolacions and resyned yourself willingly, gladly and fully for his loue to his holy protection. (More, 1534, Margaret Roper, 510) (19) the newdropping teares of somm might make you remember if it weare possible you could forgett. Quis talia fando Temperet a lachrimis? Myrmidonum Dolopumue aut duri miles Ulissei?7 and weare not I unthanckfully forgettfull, if I should not remember my noble frend (Stuard, 1603, Arabella Stuart, 158) (20) The word pauperie is meant as is shewn by the whole line Non indecora pauperie nitens to be a Life rather distant from Ostentation than in poverty, and which answers very well to my way of living, (Montagu, 1744, Mary Wortley Montagu, II,319) All the Italian quotations identified as such in our data are things Hester Lynch Piozzi reports her Italian husband either says frequently (21) or has uttered conserning a specific occasion. (21) Oh! we will part no more tho’,– per non separarti piu as he says. (Piozzi, 1784, Hester Lynch Piozzi, I,75) Proverbs are not in any significant way different from quotations, except that they lack the possibility of familiar connection of referring to the words of a family member, friend or acquaintance. There are eight passages identified as proverbial, five of them in the eighteenth century. Half are in Latin, and two each in French and Italian. The proverbs are often identified as such in the text (22). (22) Right deare brother, I find to true the Frenche adage, Qu’un mal ne vint jamais seul (Royal1, 1585, Elizabeth I, 19)

Other prefabricated expressions which show a somewhat fixed nature, in that they appear elsewhere in that form but are too short to be regarded as quotations or proverbs also occasionally appear. There are 15 of these altogether, eight in Latin and seven in French. (23) only natural affection ab incunabulis8 (Royal1, 1586, Elizabeth I, 29) (24) It was certainly a chef d’oevre of a Letter (Montagu, 1727, Mary Wortley Montagu, II,75) 4.2.3. Terminology A borderline case between code-switching and borrowing, the use of terminology originating in languages other than the base language of the text, in this case English, is yet an interesting feature of the use of multilingual resources. In most instances writers need a certain level of education to be able to use the foreign-origin elements, whether they are considered as switches or more established elements of the language. Altogether 164 instances of the material included in this study can be regarded as terminology of one sort or another. The fields included range from the law, military and politics to science, medicine and all areas of the arts. Also forms of address or parts of titles used for foreign people can be regarded as examples of terminology in a loose sense. Unlike men, women employ only a few Latin legal terms (Nurmi and Pahta 2004: 433, 439–41). Margaret Paston is one of the few women to use these terms, and her family’s constant legal battles would have made this formidable lady of the house as familiar with the appropriate names of writs and the like as the male members of the household (25). It should be noted that Latin legal terminology greatly diminishes in frequency after the fifteenth century, which most likely reflects the vernacularisation of legal practice. That is not to say that some terms can not be found later, as in (26), albeit here in a figurative use. (25) Item, the supersedias and the supplicauit is delyuerd to Alblastre and to Wechyngham, and they haue mad out bothe warantes and supersedias; neuertheles ther is non seruyd yet. (Paston, 1465, Margaret Paston, I,311) (26) Miss Lynch has arrested me with a Habeus Corpus. (Carter: 1742, Elizabeth Carter, 120) Medical terminology is surprisingly rare, considering what a frequent topic medicine and health are in correspondence (27), and that women were generally in charge of the healthcare of their families. As with legal terminology, the vernacularisation of the field may be responsible for the scarcity of latinate medical terms, and in the eighteenth century the only term found is also clearly performing double duty as a euphemism (28). (27) I prey yow þat ye wol vowchesawf to speke to Jamys Gloys to bye þe vngwentum album þat I spake to hym fore, (Paston: 1451, Margaret Paston, I,241)

(28) Dear Amelia is much better for her laxativum, as old Schwellenberg used to say. (George3A: 1798, Princess Augusta, III,501) The topic of science is rare enough in women’s letters, and the few terms found in the eighteenth century are only tentatively scientific, as in (29). Queen Elizabeth is also found employing a sole logical term (30). (29) whenever I quit Terra firma9 it is for the Regions of Air. (Carter: 1754, Elizabeth Carter, 148) (30) I protest I haue no suche in your realme, for, if the principal faile me, I shal neuer care for adiacentia. (Royal1: 1591, Elizabeth I, 64) Religion was another domain of life which was increasingly vernacular in the period we study. The only instances of what could be regarded as religious terminology, are found in the letters of a Catholic nun, Winefrid Thimelby, as she describes the course of her days (31). (31) but I hope you received my mea Culpa, wch I presently dispatched to you. For when yr father, prophet lyke, admonished me of my sin, I presently cryed, peccavi. (Thimelby: 1675, Winefred Thimelby, 85) All the arts from music to theatre provide their own terminology, and many of these usages appear at a time when the concepts are first introduced to English. While some of the words have since established themselves as part of English, it can be argued that during the early stages of their sojourn they retained a certain foreignness about them. Obviously, as soon as the words start gaining figurative meanings, it may well be that they are more or less established in the language of at least the cultured elite. (32) I hear that in England all goes on piano piano, but still it does creep on towards the opening the eyes of the world to many things they hitherto saw darkly and falsely. (Lennox: 1787, Sarah Lennox, II,60) (33) & as soon as I have read myself into a forgetfulness of my old Dramatis persona, (BurneyF: 1779, Fanny Burney, III,345) Book titles in the original language and occasional mentions of items of clothing and types of food seem to go together with other cultural contacts. Here again, these items are sometimes used in a figurative rather than literate sense. (34) I have, too, met with a french book, which pleased me – Mon bonnet de nuit (Wollstonecraft: 1787, Mary Wollstonecraft, 137) (35) they have no mark of a religious habit, being set out in their Hair and having no guimpe, but wearing des collets montez, for which I have no name in English, (Montagu: 1749, Mary Wortley Montagu, II,419) (36) You will have so many Letters to-day from Paris that mine will be a hors d’oeuvre. (Gower: 1790, Elizabeth Leveson Gower, 29)

Terms used within the highest social strata to refer to phenomena typical of their life often appear in French in the eighteenth century, when the aristocracy was blamed for speaking too much French and too little English. Particularly terms like tête à tête (37) are highly frequent in the data. Similarly, the Italian caro sposo in the letters of Hester Lynch Piozzi could be simply a reference to her Italian husband, but in the letters of the unmarried governess Agnes Porter (38) it seems to suggest that this has been a more frequently used expression, perhaps somewhat euphemistic once again. The OED (s.v. sposo) has its first quotation of sposo in 1778 and caro sposo in 1792, both from Fanny Burney, who was a friend and correspondent of Piozzi’s. By the time Jane Austen puts the expression in the mouth of a pretentious character in Emma (1815), the expression has clearly gained specific connotations.10 Similarly, references to a certain type of young man (according to the OED s.v. beaux garcon ‘a handsome fellow; a fop’, with the first quotation from 1665) are clearly meaningful descriptions of the particular social group the person discussed belonged to (39). (37) & an own postchaise took us to Ash Park, to dine tete a tete with Mr Holder, Mr Gauntlett & James Digweed; (Austen: 1800, Jane Austen, 54) (38) I have a thousand thanks to present you and Mr Talbot for your goodness to my sister and her Caro Sposo. (Porter: 1798, Agnes Porter, 198) (39) & so begune to be a littel uncivil, which I could not sufer from an uglye baux garscon. (Original3: 1678, Nell Gwynne, 25) 4.2.4 Text structuring Another function identified for code-switching in correspondence in our earlier research is text and discourse structuring. In addition to switched quotations, where the code-switch co-occurs with a shift in voice, or genre-specific formulae such as leave-takings, where the switched segment also marks the end of the text, foreignlanguage elements can be used to form lists (40) or to express shifts in topic (41–42) or changes in footing (43). There are 28 passages we have identified as having a textand discourse-structuring function in our data. These appear overwhelmingly in the eighteenth century and in French. A propos is the most common device in this usage, but also en attendant can sometimes signal the ending of one topic. A typical term signalling the confidentiality of a passage is entre nous. (40) but where I shall next proceed is not very certain. Imprimis I shall stay 3 or 4 days with my Aunt (Carter: 1741, Elizabeth Carter, 110) (41) It would not be much amiss to copy it, and send to our Friend Mr. Cadell. A propos, which Month’s review contains the Criticisme on my Book? for I cannot find it, (Piozzi: 1789, Hester Lynch Piozzi, I,328) (42) & I mean to send it on as soon as I can get a ditto, (tha means a frank,) which I hope to do in a day or two. – En attendant, You must rest satisfied with knowing that (Austen: 1800, Jane Austen, 52) (43) But that’s entre nous, for I have no notion of owning it is possible he can be spoiled to any but friends. (Lennox: 1765, Sarah Lennox, I,161)

4.2.5 Euphemism Foreign language passages in the letters often serve more than one function, and so we have already interpreted some earlier examples as being possibly also euphemistic. A clear example of euphemistic language (44) can be found in the use of doggerel Latin as a way of preventing servants from reading a passage, which is possibly also euphemistic for sexual reference (we owe this interpretation to Daybell 2004: 154–5). (44) and I sound repayment, so as when we meet, there will be pay, and repay, which will pass and repass, allgiges vltes fregnan tolles, thou knowest my mind, though thou dost not understand me. (Thynne: 1600?, Maria Thynne, 37) In addition to the avoidance of physical and sexual embarrasment, euphemisms are typically found in expressions of emotion or some type of social awkwardness. Particularly the pauvre animal in (46) in reference to the king was clearly in reference to a sensitive issue. (45) for Im sure you woud not have been so disagreable as not to have made it when you know it is a pleasure and Satisfaction to me to do any thing you desire, by which you may find you are not Sans Consequence to me, (Swift: 1732, Elizabeth Germain, III,450) (46) I don’t know what, for I must not say what a pauvre animal I think him; (Lennox: 1768, Sarah Lennox, I,218) 4.2.6 Frequent switching The frequency of switching seems to depend on many factors, including language skills of the writer and recipient. Even unskilled writers would be able to produce occasional pre-fabricated passages, whether fairly formulaic phrases such as addresses outside the letter, familiar quotations or terms for new objects or fields. The more frequent switching was, the more skill in the language(s) in question seems to be needed. Being able to produce freely formed expressions and embed them into English structures demands the kind of fluency seemingly only possessed by a few of our informants.11 Indeed, our data supports the contemporary estimation that the English aristocracy – or at least some members of it – were fluent in French, and at times seemed to prefer it to English. Even so, very few letters contain a notable frequency of switching. One of the triggering experiences seems to be a prolonged sojourn in France. Elizabeth Leveson Gower was in Paris with her husband (who had been appointed as ambassador), and her letters to her mother-in-law are pervaded with French expressions. They give an impression of a daily life lived mostly in French, so that the writer is unable to express herself completely in English anymore. In fact, one of the letters comments on this. (47) Vous vous flattez peut être Miladi, que ces Messieurs sont partis, mais je suis charmée de vous dire qu’ils sont encore ici, and are to remain a few days longer, and I hope are amusing themselves very well. I am very sorry that the time for their departure draws so near. Granville seems to be in a

very good state of health. If nobody takes him by the nose, and if he keeps rather quiet for some time, he will be very strong, as he is formé en Hercule. I think you will find his Shoulders pulled back by Mr. Huard, and perhaps his manner of coming into a room improved, but as he has naturally a good manner, les grâces d’acquit ne paraitront pas peut-être tant que s’il avait été gauche, et contrefait avant de les apprendre. Lord Garlies and I talk so much French, and in so very French a manner, that as I am just come from playing at Billiards with him you must not be surprised at my having begun my Letter in that Language. (Gower: 1790, Elizabeth Leveson Gower, 30) There are very few informants of this kind, however, and most writers who spend extended periods on the continent do not succumb to this kind of frequency of switching. This kind of linguistic allegiance to two languages could be seen as linked to Stoessel’s (2002, 93) idea of social networks as an ‘organizing mechanism underlying individuals’ choices’. It is possible that Leveson Gower’s mother-in-law was more receptive to frequent switching into French than for example Mary Wortley Montagu’s correspondents, since she was certainly highly accomplished in foreign languages. 5. CONCLUSION Women in early and late modern England employ multilingual resources available to them in much the same way their modern counterparts do. However, because of the limited educational opportunities of women in the four centuries under study here, it is only the educated elite that is able to draw on these resources. In comparison to male contemporaries, women’s range of language skills is narrower, focusing mostly on French, with the occasional appearance of Latin and Italian, both of the latter in highly conventionalized form. It is also evident that women resort to code-switching more rarely than male letter writers. The functions of switched passages range from genre-specific conventions (particularly addresses) and terminology to expressions of intertextuality (quotations and proverbs) and performing membership of the French-speaking elite of the eighteenth century. Switching seems to have been linked to both the writer’s own language skills and identity and to their correspondents’ ability to understand the languages used. Given the limited educational opportunities of women, it is remarkable how skillfully women make use of the resources available to them, often providing translations or paraphrases of more difficult passages and commenting on their choice of a particular language. There is also evidence in our data on the social propriety of knowing foreign languages and displaying that knowledge. Learning for women was not viewed favourably by all, and accomplished scholars like Elizabeth Carter could be ridiculed for their skill and learning. Similarly, knowledge of French could be seen as the appropriate skill of the aristocracy and gentry, and attempts by the lower ranks in learning the language could be mocked not only for the inadequacy displayed, but also for the insufficient familiarity with the proper social context in which to apply these skills. Since women’s place in society was much more curtailed than men’s, also their attempts at social rise and claims to learning were more limited. The study of historical data brings to light many of the same processes and strategies of multilingual communication as are present in present-day societies. In the

past, as now, multilingual literacy is connected to the availability of educational opportunities. The use of linguistic resources to express identity and membership of a group shows similar trends regardless of the timeperiod studied. In this way, historical data can highlight, in the spirit of the uniformitarian principle, the basicness of trends in multilingual communication, allowing us to view the practices of our contemporaries in new light. Our study also illustrates some benefits of using corpus-linguistic methods in code-switching research. An electronic corpus like the one we have used in this study, annotated for foreign language, enables systematic and speedy data retrieval, where the quantification of the findings according to the language-external factors coded into the corpus structure, such as the informants’ gender and education, is relatively straightforward. A more nuanced picture of the phenomenon can be obtained by combining the corpus-linguistic analysis with a qualitative analysis, where the data extracted from the corpus with the help of a computer concordancer is carefully examined in its micro-context. In our view, a full understanding of code-switching practices in a historical perspective can only be obtained by combining the quantitative and qualitative paradigms. A key concern in such analyses is the availability of data, i.e. electronic corpora that provide sufficient representative and reliable evidence for studying code-switching from the specific angle that the researcher is interested in, be it grammatical, textual, discoursal or pragmatic, or sociolinguistic, as in the present study. The Corpus of Early English Correspondence, providing the data for our analysis of gendered multilingual practices in a historical perspective, has been specifically designed to facilitate sociolinguistic research in the history of English. We hope to have shown in this study that it is possible to obtain interesting results about multilingualism and code-switching practices in a sociohistorical angle using corpus-linguistic methods, even if the corpus in question has not been compiled with the study of multilingualism in mind. Eventually, we believe, the on-going and future work in the dynamic field of corpus compilation will improve the ground for corpus-facilitated historical code-switching research with the introduction of corpora specifically designed for studying the history of multilingualism. In the meantime, the historical corpora available to date, although compiled for studying the history of individual languages, also provide a window to the history of code-switching in writing. REFERENCES Bell, N. D. 2007. Safe territory? The humorous narratives of bilingual women. Research on Language and Social Interaction 40(2–3): 199–225. Black, J. 2001. Eighteenth-century Britain 1688–1783. Basingstoke: Palgrave. Brink, J. R. 2010. Literacy and education. In A New Companion to English Renaissance Literature and Culture, Vol 1, ed. M. Hattaway, 27–37. Chichester: Wiley-Blackwell. CoRD = Corpus Resource Database. Corpora of Early English Correspondence at http://www.helsinki.fi/varieng/CoRD/corpora/CEEC/index.html. Couchman, J. and A. Crabb, eds. 2004. Women’s Letters across Europe, 1400–1700: Form and Persuasion. Aldershot: Ashgate. Cressy, D. 1975. Education in Tudor and Stuart England. (Documents of Modern History.) London: Arnold. Cressy, D. 1980. Literacy and the Social Order. Reading and Writing in Tudor and Stuart England. Cambridge: Cambridge University Press.

Daybell, J. 2004. ‘I wold wyshe my doings myght be … secret’: Privacy and the social practices of reading women’s letters in sixteenth-century England. In Couchman and Crabb eds, 143–62. Guerrero, M. D. 2003. Acquiring and participating in the use of academic Spanish: Four novice Latina bilingual education teachers’ stories. Journal of Latinos and Education 2(3): 159–81. Hawley, J. 2009. Elizabeth Carter. Oxford Dictionary of National Biography, online edition. Oxford: Oxford University Press. http://www.oxforddnb.com/view/article/4782; accessed 9 June 2010. Heller, M. ed. 1988. Codeswitching: Anthropological and Sociolinguistic Perspectives. Berlin: Mouton de Gruyter. Keen, M. 1990. English Society in the Later Middle Ages 1348–1500. (The Penguin Social History of Britain.) London: Penguin. Kraver, J. R. 1997. Revolution through poetic language: Bilingualism in Latina poetry from la Frontera. LIT:Literature, Interpretation, Theory 8: 193–206. Laurence, A. 1994. Women in England 1500–1760. A Social History. London: Phoenix Giant. Martin, J. 1998. A Governess in the Age of Jane Austen. The Journals and Letters of Agnes Porter. London: Hambledon Press. Mills, J. 2006. Talking about silence: Gender and the construction of multilingual identities. International Journal of Bilingualism 10(1): 1–16. Nevala, M. and M. Palander-Collin. 2010. ‘O England! England! She says – my Father – my Sisters – my friends! – shall I ever see you more?’: Reporting in 18th-century correspondence. In Discourses in Interaction, ed. S.-K. Tanskanen, M.-L. Helasvuo, M. Johansson and M. Raitaniemi, 133–50. Amsterdam: Benjamins. Nevalainen, T. and H. Raumolin-Brunberg. 2003. Historical Sociolinguistics. London: Longman. Nurmi, A. forthcoming. Early Modern English. In Historical Linguistics of English, ed. A. Bergs and L. Brinton. (Handbooks of Linguistics and Communication Science). Berlin: Mouton de Gruyter. Nurmi, A. and M. Nevala. 2010. The social space of an eighteenth-century governess: Modality and reference in the private letters and journals of Agnes Porter. In Social Roles and Language Practices in Late Modern English, ed. P. Pahta, M. Nevala, A. Nurmi and M. Palander-Collin, 163–89. (Pragmatics and Beyond NS 195.) Amsterdam: Benjamins. Nurmi, A., M. Nevala and M. Palander-Collin, eds. 2009. The Language of Daily Life in England (1400–1800). (Pragmatics and Beyond New Series 183.) Amsterdam: Benjamins. Nurmi, A. and P. Pahta. 2004. Social stratification and patterns of code-switching in early English letters. Multilingua 23: 417–56. Nurmi, A. and P. Pahta. 2010. Preacher, scholar, brother, friend: Social roles and code-switching in the writings of Thomas Twining. In Social Roles and Language Practices in Late Modern English, ed. P. Pahta, M. Nevala, A. Nurmi and M. Palander-Collin, 135–62. (Pragmatics and Beyond New Series 195.) Amsterdam: Benjamins. OED = Oxford English Dictionary Online. Oxford: Oxford University Press. http://dictionary.oed.com/; accessed 9 June 2010.

Pahta, P. 2004. Code-switching in medieval medical writing. In Medical and Scientific Writing in Late Medieval English, ed. I. Taavitsainen and P. Pahta, 73– 99. Cambridge: Cambridge University Press. Pahta, P 2007. Flowers, cum multis aliis quae nunc praescribere longum est: On code-switching in Early Modern English medical texts. In Tracing English through Time: Explorations in Language Variation, ed. U. Smit, S. Dollinger, J. Hüttner, U. Lutzky and G. Kaltenböck, 259–72. Vienna: Braumüller. Pahta, P. 2011. Code-switching in Early Modern English medical writing. In Medical Writing in Early Modern English, ed. I. Taavitsainen and P. Pahta, 115–32. Cambridge: Cambridge University Press. Pahta, P. and A. Nurmi. 2006. Code-switching in the Helsinki Corpus: A thousand years of multilingual practices. In Medieval English and its Heritage: Structure, Meaning and Mechanisms of Change, ed. N. Ritt, H. Schendl, C. Dalton-Puffer and D. Kastovsky, 203–20. Frankfurt: Peter Lang. Pahta, P. and A. Nurmi. 2007. What we do cón amore: Structures of code-switching in eighteenth-century letters. In ‘Of varying language and opposing creed’: New Insights into Late Modern English, ed. J. Pérez-Guerra, D. González-Álvarez, J. L. Bueno-Alonso and E. Rama-Martinez, 401–20. Bern: Peter Lang. Pahta, P. and A. Nurmi. 2009. Negotiating interpersonal identities in writing: Codeswitching practices in Charles Burney’s correspondence. In Nurmi et al. eds, 27– 52. Palander-Collin, M., M. Nevala and A. Nurmi. 2009. The language of daily life in the history of English. Studying how macro meets micro. In Nurmi et al. eds, 1–23. Pavlenko, A., A. Blackledge, I. Piller and M. Teutsch-Dwyer, eds. 2001. Multilingualism, Second Language Learning, and Gender. Berlin: Mouton de Gruyter. Porter, R. 1991. English Society in the Eighteenth Century. (The Penguin Social History of Britain.) London: Penguin. Sadiqi, F. 2008. Language and gender in Moroccan urban areas. International Journal of the Sociology of Language 190: 27–47. Schendl, H. and L. Wright, eds. forthcoming. Code-switching in the History of Earlier English. (Topics in English Linguistics.) Berlin: Mouton de Gruyter. Stoessel, S. 2002. Investigating the role of social networks in language maintenance and shift. International Journal of the Sociology of Language 153: 93–131. Valdés-Fallis, G. 1978. Code-switching among bilingual Mexican-American women: Towards an understanding of sex-related language alternation. International Journal of the Sociology of Language 17: 65–72. 1

The research presented here has been supported by the Academy of Finland, the Research Unit for Variation, Contacts and Change in English (VARIENG) and the University of Tampere. 2 In the examples in this article, expressions in languages other than English appear in italics, in the corpora they do not necessarily. In examples from CEEC400, the reference indicates collection name, year of writing, writer’s name and the page number of the edition referred to. For further information on the collections, see e.g. the appendix in Nurmi et al (2009). 3 In the compilation of CEEC400, Arja Nurmi was responsible for the ultimate decisions whether something was annotated as foreign language or not. In annotating, some attention was paid to decisions of editors in e.g. italicising lexical items, but there were other consistently used criteria. Inflectional endings were the most decisive one: when e.g. numerals were accompanied with Latin rather than English endings, they were annotated as foreign language. Names of books and titles of people were not annotated, neither were highly frequent items such as item or anno. Terms related to Christianity (pater noster, amen) were also treated as borrowings and were not annotated. The default decision in unclear cases was not to annotate. Because of our knowledge of the decisions made in annotating the corpus, we feel fairly confident in using the annotation for collecting the data, although

in the course of this study we had to revise some earlier decisions of borderline cases. Most notably, names of books and other such items have been included here as indicators of foreign language skills and interests. 4 The omitted passages consist of some longer stretches of Latin and French. Greek was also not transliterated as part of the corpus compilation process; all passages in the Greek (or other non-Roman) alphabet are marked as omitted. We have not added the word counts of the omitted passages to the total word count, but their length is not sufficient to influence the overall results. 5 Third Sunday after Easter. 6 Our thanks to the anonymous reviewer for identifying this quote. 7 ‘What Myrmidon or Dolopian soldier of harsh Odysseus could refrain from tears at the telling of such things?’ Virgil, Aeneid 2.6–8. 8 ’from the cradle’ 9 In the OED since 1693 in this sense of earth as opposed to the air. 10 Our thanks to the anonymous reviewer for the suggestion of Emma. 11 The use of the term informant here is in accordance with e.g. Nevalainen and Raumolin-Brunberg (2003, 26–52). While the subjects of research did not actively participate in the study, they were selected on similar principles as informants for a present-day sociolinguistic study.