How Women Wrote about Themselves: A Corpus-informed Comparison of Women Writers’ Defences in Sixteenth- and Seventeenth- century England

Women’s written defences of their sex developed within the literary context of the querelle des femmes, a mainly male debate on female intellectual worth, which started in the Middle Ages and came to a peak during the Renaissance. This paper focuses on the discourse some women writers started to develop in sixteenthand seventeenthcentury England, when the growing number of misogynist attacks led some of them to respond in kind. The first works directly engaging the topics of the querelle are usually identified in Isabella Whitney’s poem The Copy of a Letter (1567) and Jane Anger’s Jane Anger Her Protection For Women (1589). These are coupled with three later pamphlets framed in the so-called Swetnam debate, from the name of the misogynist pamphleteer to whom these three women writers replied, namely A Muzell for Melastomus (1617) by Rachel Speght, Ester Hath Hang’d Haman (1617) by Ester Sowernam and The Worming of a Mad Dogge: or, A Soppe for Cerberus, a Redargution of the Bayter of Women (1617) by Constantia Munda. The qualitative reading of the texts reveals differences among them both in content and structure, which is supported by a corpus-informed quantitative comparison between the earlier and the later texts. The quantitative analysis also shows differences in the use of specific high and low frequency querelle-related lemmas, which signal a variation in the semantic fields related to the discourse on women. Such a mixed research method approach suggests that these variations could not be entirely ascribed to those literary works which had established the formal guidelines of the genre of the controversy, The Praise of Folly (1509) by Erasmus of Rotterdam and De nobilitate et praecellentia foeminei sexus (1529) by Heinrich Cornelius Agrippa von Nettesheim. To understand where these differences may come from, the English querelle texts are quantitatively compared with a small corpus of Italian defences of women, including Il merito delle donne (The Worth of Women, 1600) by Moderata Fonte and La nobilità, et l’eccellenza delle donne, co’diffetti, et mancamenti, de gli huomini (The Nobility and Excellence of Women and the Defects and Vices of Men, 1600; 1621) by Lucrezia Marinella. This quantitative analysis shows that English and Italian women writers appear to share some structural and content-related characteristics which cannot be found in either Erasmus’ or Agrippa’s works or contemporary English writers dealing with the querelle. The hypothesis is thus advanced that there may have been literary contacts between England and Italy that indirectly influenced the development of the discourse on women within the context of the querelle des femmes.


Introduction
Corpus linguistics has been used more extensively in the exploration of the linguistic patterns of contemporary literary genres than in the investigation of texts from the past. The reasons for this are manifold: the generally small size of corpora based on ancient texts affects the reliability of the results of the quantitative analysis of texts because of the impossibility of "making generalisations of any worth" from a statistical point of view (McEnery, Baker 2017: 4); the occurrence of variants in spelling and inflection may also make it difficult to search for, retrieve and examine different word forms of the same terms, eventually leading to problems in lemmatization (Baron Rayson Archer 2009: 2).
Despite these difficulties, corpus linguistic tools and methods have increasingly contributed to an understanding of the workings of literary texts (see, e.g. Mahlberg 2007;Withington 2013), including ancient texts (see, e.g. McEnery, Baker 2017). Following the steps taken by these earlier studies, my research aims to use such a quantitative corpusinformed method together with qualitative analyses to investigate an early modern literary sub-genre, namely the querelle des femmes in England, also in relation to its Italian analogues.
Having originated in the classical world, the querelle des femmes is a debate over women's nature and excellence, which had been part of a wide, all-male discourse until the first female interlocutor, Christine de Pizan (1365 -ca. 1430), appeared on the literary scene. She provided a proper oppositional voice through her Letter to the God of Love (1399) and The Book of the City of Ladies (1405), turning this male narration into an actual debate. The structure of the querelle drew on the genre of the controversy, influenced by Erasmus of Rotterdam's Praise of Folly (1509) and Heinrich Cornelius Agrippa von Nettesheim's De nobilitate et praeecelentia foeminei sexus (1529) in tone and structure: these authors typically included in their works quotations and lists of biblical and historical examples to support their thesis and to deconstruct their opponents' reasoning. Despite this common formal framework, the debate developed rather differently in Italy and England.
From its classical origins onwards, Italy had been one of the most fertile terrains for the querelle. Its geopolitical fragmentation allowed the creation of many centres of power, each with its female leading figure, while its classical tradition supported at least basic education for gentlewomen. Albeit confined to minor or pious literary genres, literate women were often used as gateways to their richer and more powerful consorts. This ambivalent, and at times blatantly misogynistic, attitude towards women as mere social mediators temporarily decreased thanks to Veronica Gambara (1485Gambara ( -1550 and Vittoria Colonna (1490Colonna ( -1547. They combined their traditional roles as loving mothers and faithful widows with their political and literary influence and soon became role models for other female intellectuals, who could finally rely on literary precedents acknowledged by a male audience as well. From the time of these authors on, "a more tangible modern sisterhood" (Cox 2008: 115) started to develop, including writers such as Lucrezia Marinella Women, 1599) by Domenico Passi. In the first section of her pamphlet she focuses on women's excellence according to historical, literary and biographical examples. Then she quotes and rebuts the misogynistic opinions of Ercole and Torquato Tasso, Sperone Speroni, and, although only partially, Giovanni Boccaccio. The second section is wholly devoted to an equally lengthy and erudite disquisition on male misconduct towards women. Throughout her work, Marinella shows her involvement in the querelle by quoting not only traditional authoritative sources, but also those contemporaries of hers who joined it, such as Moderata Fonte.
Unlike Marinella, Fonte chose the genre of the dialogue for her Il merito delle Donne (The Worth of Women, 1600), which features seven Venetian women freely expressing their opinions on the female and wifely condition. Influenced by Boccaccio's Decameron, the author has them divided by Queen Adriana into two groups: one supports female excellence (Leonora, Cornelia, Corinna), the other defends male superiority (Elena, Verginia, Lucrezia). The first giornata focuses on the telling of stories about men's either malicious or virtuous nature, while the second giornata recalls a libro de secreti 1 (book of secrets), possibly a rhetorical strategy to prove female erudition also in more scientific fields. The closure interestingly mixes genres and topics, as it conflates a polemical sonnet about the mythological disappearance of True Love and a madrigal on women's growing self-awareness of their place in the world.
Lacking a heterogenous political asset and being less thoroughly influenced by humanistic sources, the English side of the debate developed differently from the Italian one. First addressed by Geoffrey Chaucer, it showed an imbalance between misogynist and proto-feminist writings, witnessing the growing popularity and effectiveness of the former. Later in the sixteenth century, women started to actively question female cultural inferiority. Margaret Tyler (ca. 1540 -ca. 1590) and Anne Bacon Cook (ca. 1528 -1610), for instance, were authors of refined translations and literary works. However, the querelle started to be explicitly addressed only by Isabella Whitney's The Copy of a Letter (1567) and Jane Anger's Jane Anger Her Protection for Women (1589). The former is a collection of verse-epistles composed in ballads where the author recalls the difficulties encountered in writing such a piece without a solid female tradition to rely on and turns to listing examples and classical sources that characterise the controversy genre. Anger's Protection for Women is the first English female-penned treatise in favour of women, written as a reply to the anonymous misogynist work Book his Surfeit in Love, a Farewell to the Follies of his own Fantasy (1588). Unlike her later colleagues' writings, Anger's work is less an encomium of womankind than a manual on how to survive men's illnature. Stylistically, it may be defined as a dramatic monologue spoken by an angry woman, featuring the traditional rhetorical forms of disputation, such as the use of allusions, sayings, and examples related to classical readings and to Book his Surfeit.
at the beginning it was generally written in Latin, and thus available only to a privileged few. (Later, the rise of the print allowed the publication of many libri in vernacular as well.) A book of secrets was a treatise which included collections of recipes and remedies for a huge range of therapeutic and cosmetic uses, sometimes including advice on household management as well. One of the most famous libri de segreti was Alessio Piemontese's (1555), which underwent more than seventy editions and numerous translations into Latin, German, Spanish and Polish by the end of the sixteenth century.
Besides these works, no other woman writer either chose the pamphlet genre or addressed the above topics in such an extensive and explicit way, preferring instead traditional genres such as religious writings, elegies, or closet dramas. This writing habit, which lasted almost thirty years, was temporarily and abruptly abandoned in 1617, when three women published proto-feminist pamphlets replying to Swetnam's The Arraignment of Lewd, Idle, Froward, and Unconstant Women (1615) The study of the socio-cultural context highlights neither an existing literary network nor a cohesive discourse on the female sex among women. Yet, these three authors seem to share a specific style and narration on gender. While never mentioning Anger or other women writers, they explicitly refer to each others' works.
Rachel Speght's A Muzell for Melastomus (1617) has been described as a "clear and recognisable starting-point from which to speak as a woman" (Martin 2010: 127). It provides a neat line of argumentation and legitimises the display of humanist education which is crucial for Swetnam's "ungodly syncretism" (Haselkorn Travitsky 1990: 50) and addressing more complicated and sensitive topics, such as Protestant theories of marriage. Her work is mentioned by Ester Sowernam, who tries to refine and problematise it in her Ester Hath Hang'd Haman (1617). Being "neither a Maid, Wife nor Widdowe, yet really all, and therefore experienced to defend all" (sig. A1r), she wittily glosses terms with foreign roots, mentions Greek authors, and updates the traditional list of excellent women of the past with some of her contemporaries, such as Queen Elisabeth I. This author mirrors Swetnam's use of common knowledge but prefers to rely on empirical observation rather than popular jokes and sayings as he did, outlining what has been defined as a "psychosocial critique of misogyny" (Haselkorn Travitsky 1990: 53). Unlike Speght, Sowernam ends her pamphlet with a mock trial, where she reverses the traditional literary trope of men talking on behalf of women by ventriloquizing her adversary, and a brief ballad called A Defence of Women, which once again summarises her claims, her attacks against Swetnam, and her advice to women in general.
While Speght's and Sowernam's works are similar in form and content, Munda's partially departs from them. Maintaining a similar argumentative structure, The Worming of a Mad Dogge (1617) employs an unexpectedly aggressive style and a lexicon that mock Swetnam's physical appearance and his gross use of language rather than countering his argumentation. Due to such a peculiar attitude, some scholars doubted Munda's gender and suggested a male identity (Haselkorn Travitsky 1990: 60). Munda's education and literary pose, characterised by sexual and scatological discourses, technical jargon, and references to ancient Greek and contemporary writers, were far more befitting of an early modern male writer than a woman. This choice might have served misogynistic purposes: crafting such a female persona might have either created literary uproar or confirmed the negative stereotype regarding educated and talkative women, eventually legitimising the violence generally used against them.
Given such a context, it is quite surprising to notice how effective and at times bold these later pamphlets are without a widespread literary tradition to rely on. Whitney and Anger's works could have acted as starting points, but they could not constitute solid literary grounds on which Speght, Sowernam, and Munda might have felt entitled to build their claims. The scarce critical attention paid to this topic has led me to formulate my first research questions: Are there actual differences in the development of the content and style within the English discourse on women? If so, do these differences derive from influences located within the English literary framework? If not, is it possible that the English pamphlets published in 1617 somehow echoed the Italian discourse on women, which not only relied on a solid tradition, but was also developed by women, such as Colonna and Gambara, and male writers who were well-known abroad? Could the English women writers publishing in 1617 have taken as their starting point their earlier Italian colleagues' works and consequently refined and adapted them to create a more effective narration and properly enter such an international debate? Finally, from a methodological point of view, can the above questions be explored through a mixed-method approach that combines qualitative and quantitative approaches in its investigation of a specific early modern literary sub-genre? If so, can this query be effective if focused on lowfrequency words as well?
The rest of the paper will address the above questions. Section 2 will deal with the method, explaining how texts were collected, divided into corpora and sub-corpora, and analysed thanks to qualitative readings and specific quantitative tools. Section 3 will present and discuss the findings. Particular attention will be addressed to the role such a mixed-method approach to the analysis of literary texts can play in checking the validity of research hypotheses. Section 4 will draw the conclusions.

Method
This paper illustrates the application of qualitative and quantitative analyses. The research first started with an extensive qualitative reading on the querelle des femmes, primarily focused on the English context (Demers 2005;Kelly-Gadol 1982;King 1991). This led to the identification of a restricted historical period within which the English discourse on women started to change both in content and in style. The period ranges from 1567 (publication of Isabella Whitney's A Copy of a Letter to an Unconstant Lover) to 1617 (the 'Swetnam's debate').
Stemming from these considerations, a corpus containing five English women writers' works has been assembled, which henceforth will be referred to as EWW (English Women Writers). This has been divided into two sub-corpora: the one including Whitney's and Anger's works will be referred to as ENG1560 and counts 10,202 tokens (7,226 tokens in Anger, 2,976 in Whitney), while the one including Speght's, Sowernam's, and Munda's works will be referred to as ENG1617 and counts 34,906 tokens (9,983 in Speght, 14,861 in Sowernam,10,062 in Munda). The corpus thus includes only those texts that are strictly relevant to the debate. The EWW corpus was compared against two well-known corpora: Women Writers Online (WWO; https://www.wwp.northeastern.edu/) and Early English Books Online (EEBO; https://quod.lib.umich.edu/e/eebogroup/) Phase II. WWO is a full-text collection of English women writers' texts ranging from 1526 to 1885 and compiled by Northeastern University. This corpus of 416 literary works can be searched by genre, date, author, title and keyword. EEBO is a well-known collection of literary facsimiles that currently includes more than 132,600 Early Modern English printed texts from 1473 to 1700. In cooperation with Text Creation Partnership (TCP), EEBO compilers created EEBO-TCP, currently a fully digitalised corpus of more than 35,000 texts. Thanks to its refined search options, it is possible to search the corpus not only by author, title, genre, and keywords, but also by decade.
A corpus of texts by Italian authors was compiled to analyse the Italian side of the debate. The corpus, which will be referred to as ITA1600, included Marinella's and Fonte's texts and counts 174,781 tokens (108,127 in Marinella,66,654 in Fonte). The reason for compiling such a corpus was that it was impossible to find a comprehensive online Italian corpus comparable to EEBO in size and search options. For example, Biblioteca Italiana (http://www.bibliotecaitaliana.it), the largest and most reliable corpus available on Italian literature, grants free access to only 3,423 texts written between the tenth and the twentieth centuries, and has limited search options.
The preliminary step of the analysis was the identification of "cultural key words" (Stubbs 1996: 172;Rayson 2008: 524), that is, "words which capture important social and political facts about a community" and keywords, that is, "words which show a change in frequency" which are "not words of high frequency necessarily" (McEnery, Baker 2017: 158), through a mixed-method approach. Cultural keywords were detected qualitatively thanks to a careful close reading of the texts. They include words such as "excellence", "worth", and "defence": a close reading of the texts reveals they carry a specific meaning when related to the discourse on women in this particular sociohistorical period. Keywords, instead, were identified thanks to AntConc (https://www.laurenceanthony.net/software/antconc/), a freeware corpus analysis toolkit for concordancing and text analysis developed by Anthony Laurence, which made possible the investigation of frequency word lists of the texts under examination.
The bottom-up lexical analysis of the texts was based on a compilation of a specific type of lemma lists, which group together inflectional variants of a given word as well as its derivatives and words very close in meaning under the same "term/concept". For instance, the lemma I includes words such as I, me, mee, my, mine, myne, and the lemma WOMAN includes inflectional variants, spelling variations, derivates, and words closely related in meaning, such as womans, women, womens, womanhood, womankind, womankinde, gentlewoman, gentlewomen, gentle-woman, gentle-women. Lemmas are written in small capitals.
In retrieving word frequency lists, a stop list was used that included several function words except personal pronouns and possessives. The reason for excluding these words from the stop list was that they were pivotal in questioning the existence of differences in use of gender references in querelle-related texts.
Wordlist analysis flags points of interest according to frequency changes in high frequency words, otherwise generally stable (Sinclair 1991: 31). However, low frequency words can also signal peculiar features in the discursive development: the rarity effect (Fortier 2002: 202) provides the reader with "stylistically-marked cohesion breaks" (Emmott 2002: 101) that consequently alert their attention. Thus, it was pivotal for this research to search for low frequency words as well. I first of all qualitatively identified candidate low frequency words, checking their spelling variants. Then I relied on the AntConc concordance tool to investigate the actual frequency of these words and compile reliable word lists. In order to make low frequency word lists comparable to the highfrequency ones automatically generated by AntConc, I normalised absolute frequencies turning them into relative ones.
Starting from these word lists, I identified groups of words relevant to the same semantic fields, that is, tightly related in meaning (McEnery, Baker 2017: 173). They were divided into "gender reference", which was then sub-divided into "male gender" and "female gender", "writer-reader relationship", and "literary awareness". Since they derive from observation of both automatic and qualitative word lists, these semantic fields are made up of both high frequency and low frequency words. Thus, for instance, "female gender" includes both those words linked to WOMAN thanks to lemma lists (womans, women, womens, gentlewoman, etc.) and those selected qualitatively as low frequency words (lady and virgin). To explore the differences between ENG1560 and ENG1617, and then between both of them against ITA1600, I compared and contrasted the frequency values of the words occurring in them, as relevant to the lemmas considered. To check the significance of the different frequency values, I employed the log-likelihood ratio test (henceforth LL) (Rayson 2008: 527). LL includes normalisation as part of the value formula and determines whether a relation between two words is casual: if the critical value is greater than 15.13 (p < 0.0001), then the probability of the result or change happening by chance is less than 99.99%, if it is greater than 6.63 (p < 0.01), the probability is less than 1%, if it is 3.84 or more (p < 0.05), the probability is less than 5%. In the tables in Section 3, where the findings are illustrated, p values are reported, followed by LL values in brackets.

Results and Discussion
The first goal of the research was to establish whether the discourse on women employed by the women writers of the 16 th and 17 th centuries belonged to a widespread English narrative. To do so, words that encode concepts particularly relevant to the querelle were investigated in EEBO and their frequency of occurrence was determined.
Relying on a qualitative reading of the texts, three "cultural key words" were chosen, namely "excellence", "worth", and "defence". These best summarise the core topics of the querelle, have a fairly broad meaning in contexts unrelated to the debate, and are hypothetically fit for many other kinds of discourses popular in Renaissance England such as those dealing with historical accounts or religious topics. Occurrences of these words retrieved in the EWW corpus were collected in the section of the EEBO corpus relevant to the same time period as the EWW corpus (i.e. 1560 -1620). This section of EEBO contains 3,445 works published in that period, with exception made for Speght's, whose work is currently not available in the corpus. Frequency counts of the above cultural keywords were computed and their collocates examined. The searches for the terms were based on their truncated forms (e.g. "excellen*") so as to retrieve all inflectional variants and derivatives of the cultural keyword, as a form of "extended lemmas". In the tables below, frequency counts will be shown as relevant to these extended lemmas.
Relative frequencies (normalised per 100 words) were unsurprisingly high: "excellen*" reached 69.52, "worth*" 70.97, and "defen*" 69.11. This meant that such words were commonly employed during this period in many literary genres. To understand whether their use was widespread with relevance to the debate on women too, I also searched for "excellen*" and the other words next to the collocate "wom*", which stands for both woman and women. Results changed considerably: "excellen*" reached a frequency value of only 2.64, "worth*" 3.72, and "defen*" 2.00. It should be pointed out that in many cases these terms are not used with relevance to the debate. For example, Thomas Harding, in A reioindre to M.Iewels replie against the sacrifice of the Masse (1567), uses "excellen*" and "wom*" to comment on a discourse far from the one of the querelle, specifically to strongly remark his spite towards bishops or popes who commit mortal sins ("this Priesthode sometimes is worthier, and of more excellencie in a woman, or a childe, then in a Bishop, yea perhaps then in the Pope him selfe. For in him it is none at al, if he happe to fal into mortal sinne", p.242). Women are also mentioned in historical accounts and usually praised for their roles of housekeepers and loving mothers; still, only a few works properly deal with the debate topic. If such "noise" (e.g. occurrences of "wom*" not relevant to the debate) is excluded, the frequency values drop to 0.15, 0.17, and 0.29, respectively.
This preliminary analysis provides support for the hypothesis that English discourse on women belongs to a specific literary sub-genre where common words, framed in such a peculiar cultural setting, gain new meanings. Sowernam, for instance, employs "defen*" to refer both to Swetnam's accusations and her writing as a literary weapon ("I am not onely provoked by this Authour to defend women" sig. B1r; "[n]ow albeit I have undertaken the defence of women, and may in that respect be favoured, in taking all advantages I may, to defend my sexe." sig. B4v). Anthony Munday, in Zelauto (1580), lends his voice to one of the protagonists, who, commenting on a poem, says: "And certainely it amazeth me to heare that such excellencie should remayne in a woman. But I pray you procéede, and let me heare more of this matter?" (page 25).
The search for the "cultural key words" "excellen*", "defen*", and "worth*" next to the collocate "wom*" was further refined to investigate the EEBO corpus across the various decades within the 1560-1620 time span (see Table 1). The findings show that, in association with "woman", the frequency of use of each of the three words generally increases over time, and reaches its peak in the same decade as in the EWW corpus, namely that from 1610 to 1619. This result is in line with the impression previously emerged from the qualitative analysis, according to which the discourse on the querelle is very specific and only minimally addressed explicitly until 'Swetnam's debate' in 1617. Table 1. Number of matches for pairs of "lemmas" (and number of works where they appear) in the EEBO corpus (1560 -1620) analysed per decade. The lowest and the highest values, relevant to the initial and final period, are given in bold.

(29) 79 (48)
The ascending trend suggests that writers varied the use and frequency of those querelle-related word combinations from one decade to the other, highlighting the presence of internal differences in the shaping of this discourse. This increase in frequency of use frames these works into more topic-specific literary scenarios and suggests that these women writers possibly became involved in a discourse specifically focused on the shaping of women's image and eventually assumed an argumentative rhetorical stance opposing the predominant male narrative. This finding and its tentative interpretation may provide a partial answer to the first research question about whether there could be actual differences in the development of content and style within the English discourse on women. To test the above mentioned hypothesis, the EWW corpus was analysed chronologically in order to identify possible differences in structure and content in the works of writers belonging to different decades, as specified in Table 1. 2 Thus, EWW five literary works were assigned to two sub-corpora according to their date of composition (ENG1560 included Whitney's and Anger's works, while ENG1617 included Speght's, Sowernam's and Munda's works). Frequency word counts were computed for both sub-corpora and the results presented under relevant lemma headings as explained in Section 2. The use of the stop list in extracting frequency counts made it possible to explore the frequency of use of personal pronouns and possessives as well. Table 2 shows the results. The data show that personal pronouns and possessives always appear in the top positions, which suggests that the focus of the texts is on people's actions and experiences. An overview of the main 20 collocates of YOU and HE shows that, besides appearing together with the, your, and his, these two lemmas co-occur with I ("I charge you with blasphemie, with impudence, scurilitie, foolery, and the like.", Sowernam, sig. G4r, my italics) and women ("Hee writeth a Booke, an Arraignment he calleth, In which against women he currishly bawleth.", Sowernam,sig. G4v,my italics). Such a use suggests that particular attention is paid to the polemical relationship between the writer and her literary opponent, and consequently hints at a growing emphasis on women's self-representation and self-expression in writing. This focus on women's self-representation is also evidenced in the dispersion of high frequency words, which creates cohesive chains through which authors link their main topics throughout their texts (Emmott 2002). For instance, the lemma I does not occur the most frequently at the beginning of the pamphlets, where the authors introduce themselves. In fact, as shown by the Concordancer Plot in AntConc, it permeates the whole text at almost regular intervals, matching the overall qualitative impression of the need for a constant affirmation of women's role as both subjects and readers. Sowernam uses the first-person pronoun to make her stance clear from the beginning ("The Author of the Arraignment, and my selfe, in our labours doe altogether disagree; he raileth without cause, I defend upon direct proofe." sig. A3v, my italics) and reasserts it at the end of her pamphlet ("I shew just and direct proofe for what I say; it is not my desire to speake so much, it is your desert to provoke me upon just cause so farre; [...] the report of the truth is never to be blamed, the deserver of such a report, deserveth the shame." sig. G4r, my italics). Anger also uses it to display her literary role, coupling it with the awareness of being both the active, authorial subject, and the passive object of this narrative ("I marvel how we women can abide them but that they delude us." sig. C4r, my italics). Similar observations apply to such "extended lemmas" as WOMAN. Table 3 shows an increase in the use of some lemmas from the first to the second sub-corpora (i.e. 1560 -1620). Other lemmas instead, such as PAGE, appear only in the ENG1617 sub-corpus. It may be claimed that these three authors borrowed such terms from one another (e.g. Sowernam from Speght, Spegth from Swetnam). Still, it may also be hypothesised that they may have either adapted those lemmas from other discourses and literary genres or absorbed them from external sources dealing with the same topics. Indeed, it seems quite unlikely that three authors, with apparently no direct contact with one another, used the same word, which also happens to constitute an innovation in the discourse they are referring to, without relying on external sources. Some differences, however, are to be signalled as well. In Table 2, in ENG1560, for instance, WOMEN ranks in a low position, appearing even after DO and LOVE, while WIFE is not even present and shows one of the lowest values (0.56). The lemma WRITE is excluded from the ENG1560 word list but included in ENG1617, where it plays a key role in the shaping of both Speght's, Sowernam's and Munda's writings and, most importantly, their idea of woman.
Since the awareness of one's literary worth clearly emerged from the qualitative reading of the texts, the decision was made to more deeply investigate the semantic field of 'literary awareness'. To this end, terms were identified through repeated readings relevant to the discourse of reading and writing. These are: PAGE, LINE, ANSWER, REPLY, BOOK, WORK, PEN, TONGUE, WORD, DEFENCE, READ. Their frequency values were computed as reported in Table 3. Table 3. Frequency word lists related to ENG1560 and ENG1617 with their respective relative frequency (per hundred words) and p values (LL) for lemmas selected manually and linked to the semantic field of reading and writing. Words whose frequency value differs remarkably from one sub-corpus to the other appear in bold.  Table 3 shows a constant increase in the relative frequency and thus use of each of the above-mentioned and manually selected "cultural key words" related to the semantic field of reading and writing. The ones whose p value is greater than 0.05 and thus bear statistic relevance are WORK, LINE, and PAGE.

Word
The last step of this analysis aimed at comparing the relative prominence of other semantic fields linked to the querelle and investigating the nature of their relation in the two English sub-corpora. To do so, I relied on ENG1560 and ENG1617 word lists and divided the lemmas showing a p value higher than 0.01 (LL value < 6.63) (see Table  4) into three semantic fields: "male gender" and "female gender" (later included in "gender reference"), "writer -reader relation" and "literary awareness". Table 4 reveals which querelle-related terms, among those whose p value is higher than 0.01 (LL value < 6.63), witness an increase or decrease in use from ENG1560 to ENG1617. The lemmas which witnessed a decrease in use from ENG1560 to ENG1617 are HE, THEY, I, WE, AND LOVE, while those which witnessed an increase in use from ENG1560 to ENG1617 are YOU, WOMAN, SHE, GOD, SAY, SELF, and WIFE. These differences suggest that women writers gradually left aside abstract discourses, such as those gravitating around "love" and definition of self, (I and WE), and preferred more authoritative narratives, such as that about religion (GOD) and personal terms to identify oneself, as YOU (generally referred to their male attacker) and WIFE. Also, the decrease in frequency of use of I and WE and the complementary increase in frequency of use of SHE, WOMAN, and WIFE suggest that ENG1560 writers felt a compelling need of stressing their presence as authors in their writings through the continuous use of the first personal singular pronoun as well as their belonging to a group of people sharing a common view of the world through the use of the first personal plural pronoun. In the later ENG1617, women writers seem to have acquired that kind of self-awareness and sense of belonging. In their writings, the discourse on women was used not as a literary tool for self-recognition, but as an issue to be seriously addressed. The use of SHE, WOMAN, and WIFE confirms this reading, since these terms place woman as the object of narration outside of the personal sphere of the author's identityunlike I and WE-and frame it within a specific social dynamic, that of marriage and, consequently, of family. Conversely, yet coherently, men are more directly addressed: while HE, and THEY tended to place some distance between them and female writers, "you" opens up the dialogical relationship and clarifies the debate-like nature and intent of these writings.
The changes in the frequency of use of words relevant to gender (I, HE, SELF, SHE, THEY, WE, WOMAN, WIFE, YOU), writer-reader relation (I, YOU) and literary awareness (SAY) and other kind of discourses (GOD, LOVE) appear to highlight a different perception of the female writers' identity and attitude towards writing. The shift from a distant and rigid division between man and woman leaves room for a more engaging and challenging relation where the addressee is treated as almost the writer's equal (YOU), whose self-awareness as an active and legitimate speaker is suggested by the increased use of SAY and SELF, and that of PAGE, LINE, and WORK, derived from the "cultural key words" in Table 3. To provide a more comprehensive picture of the above-mentioned changes in word frequency use, I grouped together words belonging to three general notions, namely "gender reference", "writer-reader relation", and "literary awareness", and computed their frequency of use. I included in each specific semantic field those keywords with a p value greater than 0.01 (LL value < 6.63) and those lemmas which showed low frequency values but were tightly bound to these semantic fields. Thus, the "male gender" semantic field includes high frequency terms such as you, your, his, he, man, men, mens, mans, mankind, him, thee, thou, thy, himselfe, hee, husband as well as low frequency ones, such as GENTLEMAN and BOY; the "female gender" semantic field includes high frequency terms, such as I, women, woman, her, she, shee, our, my, wife, we, wee, us, me, mine as well as low frequency ones, such as GENTLEWOMAN, LADY, and VIRGIN; the "relation writer -reader" semantic field includes high frequency terms, such as I, YOU, THEY, to which the low frequency term READER was added; while the "literary awareness" semantic field includes high frequency terms, such as SAY, WORK, LINE, and PAGE, to which low-frequency lemmas, such as WRITE, READ, ANSWER, BOOK, TONGUE, PEN, WORD, REPLY, and SURFEIT were added. Table 5 illustrates the frequency values of the terms belonging to the above-mentioned semantic fields in ENG1560 and ENG1617 with their p values. It shows that over time women writers appear to attribute more importance to their literary self-affirmation. Given that a p value higher than 0.01 points to a 99% certainty that the finding obtained is not due to pure chance, these data suggest the discourse of the querelle witnessed a harmonious development in England. In under fifty years, texts representative of female literary engagement in this peculiar literary sub-genre show a significant increase in the attention devoted to the core discourses of the querelle itself. Moreover, these data seem to indicate a decrease in the focus on gender binarism within the context of a conceptual framework which prioritises literary self-affirmation. According to the results, the "gender reference" semantic area (p value < 0.05) progressively leaves the stage to the metaliterary dimension (p value < 0.0001), which aimed not only to create a fictional debate but also to introduce within a rhetorical context a less stigmatised discussion about popular issues and was thus key to the controversy genre. These results appear to be in line with the differences in the frequency of use of specific querelle-related terms reported in Table 4: in that case as well, the lemmas showing a higher p value were the ones which possibly pointed to women's more active, dialogical attitude towards their male counterpart (YOU, SHE) rather than a reflexive, meditative pose regarding their identity (THEY, WE).
On the other hand, the findings relevant to the "writer-reader relation" semantic field are somewhat surprising. Since the ENG1617 corpus included pamphlets only, I obviously expected to see its dialogic dimension-as evidenced by the use of first and second-person pronoun-to be more prominent than that of ENG1560 corpus, which includes only one pamphlet and then a poem as well. Instead, the frequency values related to the "writer-reader relation" are higher in ENG1560 than in ENG1617. This result hints at a stylistic trend reversal in women writers' refashioning of the literary genre of the querelle. Moreover, it may also be noticed that the use of PAGE, LINE, and BOOK is mainly, if not exclusively, present in ENG1617. This lends support to the impression formed from the qualitative analysis that quoting sources is a feature of later pamphlets only and is missing from their English literary antecedents. Anger, for instance, makes use of specific references; yet she does not mention lines, chapters, or books (" [...] at the latter end of his book affirmeth, that already he half repenteth of his bargaine [...]", sig. D1r).
At this point, an objection could be made: because of their authoritativeness and foundational role in the controversy genre, Erasmus's and Agrippa's works may have influenced Speght's, Sowernam's, and Munda's structure and style. These three authors could have come in touch with Agrippa's and Erasmus's works due to their widespread circulation. Erasmus was translated both by D. Clapham in 1542 and by W. Bercher in 1559 (Van Der Poel 1997:87), while Agrippa "was frequently mentioned in English publications, including texts by John Donne [...] and Thomas Nashe", thus making them, at least from a printer's perspective, popular authors (Dodds 1999:146). Yet, Agrippa never employs the above-mentioned lemmas to quote his sources and Erasmus often refers to the works he mentions in loose terms ("Peter received heaven keies: yea received theim at his handes (saie they) that woulde never have committed the same to one unwoorthie theim.", The Praise of Folly, 52). As the AntConc Concordancer Plot shows, in The Praise of Folly, CHAPTER appears eight times and is used in one section only (few times ("And as he [Saint Paul] understood charity well himself, so he did as illogically divide and define it to others in his first Epistle to the Corinthians, Chapter the thirteenth.", The Praise of Folly, 52). Unlike Agrippa and Erasmus, Sowernam and Speght in particular employ these lemmas more coherently. Sometimes, they use them to refer to authoritative sources: […] as hereafter shall be shewed, maugre the shamefull, blasphemous and prophane speach of Joseph Swetnam, page 31. beginning line 15. as followeth. (Esther Hath Hang'd Haman;C1v) To sum up the findings so far, the quantitative analyses support the qualitative impression that in under fifty years English women writers developed their own discourse on female sex in a harmonious and cohesive way, differentiating it from broader narrations which made use of the same keywords. Michael Drayton's description of women as moral enemies to men ("that high and mighty Lord, his people doth defend, / And by a silly womans hand, hath brought him to his end.") greatly differs from Speght's far more critical account: Woman sinned, it is true, by her infidelitie in not beleeving the Word of God, but giving credite to Sathans faire promises, [...]  Moreover, these analyses also point to differences in the use of specific high-and low-frequency lemmas within the English discourse. Anger, for instance, does not employ those low-frequency lemmas (PAGE, LINE, CHAPTER) used by Speght and Sowernam to refer to authoritative sources and Swetnam's work. Although she replies to a misogynist pamphlet too, thus hypothetically complying with the same rule dictated by such a literary controversy, Anger refers to her sources only in general terms and mentions her opponent's work as "the surfeit" only ("What Nature hath made, Art cannot marre, (and as this surfeiting lover saith) that which is bred in the bone, will not be brought out of the flesh", B4v-C1r). These variations seem to suggest an internal change in the English discourse on women, in particular for what concerns the attention given to some semantic, and thus conceptual, areas, as in the case of "gender reference", "writer -reader relation", and "literary prowess". This change happens in one specific year-1617-with no apparent derivation from a prior English literary influence.
To verify the hypothesis that there may not have been a female literary network for English women writers, I carried out a further investigation on the WWO corpus by means of close reading of the texts published between 1560 and 1620. The analysis of the 34 texts thus retrieved suggests that the majority of them could barely have had an impact on Speght's, Sowernam's, and Munda's writings in terms of style and content. The reasons for this are that they belonged to very different literary genres and, most importantly, lacked those stylistic features that characterise the controversy genre. At most, I think they may have worked against the common stereotype of women's inferiority, proving their worth in traditional male fields such as politics and 'high' literature and contributing to shaping a gradually new idea of femininity. Elizabeth I's speeches, such as The Tilbury speech (Aske's version) (1588) and The Golden Speech (1601) (1587) consists of a translation of the chivalric romance Espejo de príncipes y cavalleros by Diego Ortuñez de Calahorra; it kindled controversies both because it stepped into a field that had thus far been considered a male prerogative and because it belonged to Continental romantic fiction, a genre generally regarded as inappropriate to women in Renaissance England. In this view, Tyler's work may be considered very courageous, as it contradicts a whole patriarchal system of values by claiming gender equality in the literary field and, implicitly, in everyday life. Tyler supports the choice of her topic by reminding the reader that women had never been completely unrelated to war stories: she recalls the mythological figures of the Amazons and points out that women had always read secular books written by men, and thus that it would be nonsensical to ask them not to write about those very same topics. Tyler also sharply criticises the institution of marriage and violence against women as common practise. While her work deals with some of the querelle topics and may have contributed to their circulation, from a stylistic point of view Tyler employs an initial deferential tone due to the commercial nature of her work, which is not to be found in relation to a male audience in the ENG1617 corpus and lacks those rhetoric strategies which, on the contrary, characterise it.
Thus, it may be concluded that, although the works of women writers published between 1560 and 1620 partially dealt with the issue on women's excellence, they did not display those specific features which characterise all of the three works published in 1617. Where did these new literary variations come from then? I carried out some other qualitative analysis on the development of the querelle in Europe and on cultural contacts between England and foreign countries prior to 1617, which suggested investigating the presence of such features in the more ancient Italian discourse.

Sub-corpus Relative frequency -"gender reference"
Relative frequency -"writer-reader relation" Relative frequency -"literary awareness"  Table 6 shows some interesting results, especially regarding the "writer-reader relation" semantic field. Although the ITA1600 corpus includes both a pamphlet and a dialogue, it shows the lowest relative frequency value for such a field (which slightly increases if the names of the characters in Fonte's dialogue are included, reaching 2.25); relative frequency value of the "writer-reader relation" semantic field peaks in the only corpus that contains the less dialogical literary genre, namely Whitney's poem. Moreover, if compared to the values of the other semantic fields, the "writer-reader relation" is the least frequently employed in the Italian corpus, while it shows medium-high values in the English ones. The reason for these variations may depend on specific stylistic changes occurring in the development of the English and Italian discourses on women, which may have signalled the need to stress the dialogical semantic field less often, as it was already intrinsic in their nature. Values concerning "gender reference" are of interest if related to those of the "literary awareness" field. ITA1600 and ENG1617 relative frequency values for "gender reference" are quite similar to each other so that, if the hypothesis of Italian echoes on the English discourse is taken into consideration, the significant gap between ENG1560 (0.97) and ENG1617 (1.53) could hardly have occurred by chance. Taking for granted that the English narration on women was undergoing an autonomous process of reshaping, albeit supported by little literary evidence, the echoes of the Italian querelle may have helped carry out structural changes and channel literary efforts on topics that needed a further development. In the Italian production, stressing gender differences was still a core topic in the debate; still, it was progressively matched by the need to assert women's literary excellence, as the ITA1600 values show. For instance, both Giuseppe Passi and Lucrezia Marinella made use of those low-frequency lemmas, such as LIBRO, to signal references from authoritative sources. This argumentative model appears to have met querelle writers' needs as it was employed in the following decades by other women writers, such as Bianca Naldi (Risposta, 1614) and Isabella Sori (Panegirico, 1628). To English women, who strongly highlighted the issue of gender imbalance, underlying their literary role may have seemed a new point of interest towards which to channel their literary efforts, probably starting from Whitney's and Anger's first attempts. However, on the English side of the querelle, the above-mentioned lemmas had not been used since John Knox first employed them in his The First Blast of the Trumpet against the Monstruous Regiment of Women (1558). They reappear in the Swetnam debate, approximately fifteen years after the exchange between Passi and Marinella (1599 -1600) and then disappear from the English argumentative model related to the querelle. These lemmas were employed again only in the 1640s by Katherine Chidley and Elizabeth Lilburne who, however, dealt with religious and political topics unrelated to the debate. This does not necessarily suggest that English women writers composing their tracts in 1617 were directly influenced by literary echoes of their Italian colleagues. Rather, these results indicate that English women writers might have relied on material other than the domestic one to develop their discourse, and that literary choices shared by both English and Italian writers may hint at the existence of a common representational framework they could rely on, which I suggest was rooted in the Italian tradition.

Conclusions
From the reading of the five English texts analysed here, I have the impression that the sudden and harmonious development of English women writers' discourse on the female sex attested to in 1617 might somehow have reflected external, foreign literary echoes, and I thus hypothesize that the more ancient and well-structured Italian tradition might have had a distant influence on them.
To test this hypothesis, I conducted a mixed-method analysis of seven texts (five written in English, two in Italian) related to the querelle des femmes composed between 1567 and 1617, which involved both close readings and data-driven analysis of lexis. The comparison of data deriving from these quantitative and qualitative analyses made it possible to discover and interpret similarities and differences in these corpora. The comparative findings obtained suggest that the changes in content and style within the English discourse on the querelle on behalf of women writers could have hardly happened by chance, and that they were probably framed within the coherent development of such a literary genre. To explore the possible source of such changes, the WWO corpus was considered: it provided material for qualitative readings and pointed to the lack of a sufficient number of women writers' works specifically characterised by such features which could have influenced ENG1617. A further quantitative analysis on ITA1600 showed that its relative frequency value linked to the "literary awareness" semantic area may help explain the sudden increase in use of this field on behalf of English women writers publishing in 1617, who had almost no literary support to carry out such a change in the focus of their narrative.
However, the hypothesis of an Italian indirect literary influence on English women writers' production relevant to the querelle is only initially tested in this paper. Future research could help trace a possible relation between the two sides of the debate, expanding the number of relevant texts to be analysed that illustrate female production in Renaissance Italy. A good starting point may be the Italian Women Writers corpus (https://www.lib.uchicago.edu/efts/IWW/), compiled by the University of Chicago, although at the moment it does not include some decisive works such as Fonte's The Worth of Women. Moreover, it may also be useful to look at other instances of cultural contacts among English and Italian writers engaging in the debate. Alessandro Piccolomini may be a pivotal character in this research: he entered the debate by writing the well-known Raffaella, o Dialogo della bella creanza (1538), which was also translated into French as Instruction pour les jeunes dames by a woman writer, Marie de Romieu (1597) (Plastina 2006: 81), and was also well-known abroad because of his comedy Gl'Ingannati (1531), which was so popular that it is usually listed as one of Shakespeare's secondary sources for Twelfth Night (1601 -1602) (Muir 1977: 132). These international literary exchanges and contacts may have facilitated the circulation of the discourse on women, carrying throughout Europe not only its popular male writers, but also its less well-known female ones.
Finally, also when working with larger corpora, qualitative and quantitative methods combined can provide useful, complementary insights into the meaning and value of literary texts, bearing in mind that also low-frequency or rare words may be crucial towards an accurate interpretation of literary discourses. Finally, additional specific tools could be employed so as to aim at more reliable stylistic and linguistic analyses. For example, Latent Semantic Analysis may be useful in analysing semantic relations among the corpora, since it consists of an investigation of co-occurrences among words resulting in a specific set of "collocations,", which are identified according to the strength of their conceptual association. Topic Modelling may be also used to automatically identify topics and derive hidden patterns from a text corpus. These analyses, together with those focused on occurrences and frequency, may lead to more reliable results and help identify discourse patterns across languages.