Introduction

‘Welsh English [WelE]’ is defined as ‘a cover term for diverse accents and dialects used in Wales’ by Heli Paulasto, Robert Penhallurick, and Benjamin Jones.¹ According to Jones, ‘the first novelist who used [Welsh English] literary dialect was T. J. L. Prichard in The Adventures and Vagaries of Twm Shon Catti (1828)’. Indeed Jones, citing Roland Mathias and Stephen Knight, writes, ‘not only was it the first to feature [Welsh English], but Prichard’s Twm Shon Catti was the first English language novel of Welsh origin to be published’.² In this article I explore the considerably earlier use of WelE literary dialect and consistent use of free indirect speech (FIS) in the 1794 novel Ellen, Countess of Castle Howel by Anna Maria Bennett, who is considered by many sources to be a Welsh author, and who came from a merchant/working-class upbringing.³ That this much earlier offering contains WelE and FIS, and comes from such an author, recontextualizes and resituates now canonical writers like Jane Austen and other historically important authors, such as T. J. L. Prichard, within a larger and earlier tradition of novel-writing style.

The two features, FIS and WelE, occur both together and independently throughout Ellen. In this article I analyse first how the FIS functions as a segue or break from narrative form before considering the WelE, which most often appears as marked speech. Daniel Gunn writes that ‘Jane Austen is generally acknowledged to be the first English novelist to make sustained use of free indirect discourse [FID] in the representation of figural speech and thought’;⁴ but, more recently, Monika Fludernik argues that ‘the last two decades of research [have] demonstrated that several earlier and some contemporaneous authors also used both FIS and FIT [free indirect thought].’⁵ Bennett is one such author. She writes with FIS, and was widely published and read, several years prior to the composition of the first drafts of Austen’s Pride and Prejudice (pub. 1813), or Sense and Sensibility (pub. 1811). Bennett may be one of the first writers to use both WelE and FIS together in the novel form. Fludernik describes FID as ‘related to: (1) syntax, (2) referential aspects (pronominal reference, tense), and (3) expressivity markers’ in texts.⁶ According to Roy Pascal, in FIS the ‘narrator […] preserv[es] the authorial mode […] [by] reporting the words or thoughts of a character, directly [from] the experiential field of the character, and adopt[ing] the latter’s perspective in regard to both time and place’.⁷ In Ellen Bennett uses WelE literary dialect both in the marked speech and in the FIS of the Welsh servant character Winifred Griffiths. Graham Shorrocks defines literary dialect as ‘the representation of non-standard speech in literature that is otherwise written in standard English […] and aimed at a general readership’.⁸ In addition to the literary dialect and FIS, Bennett provides a metalinguistic commentary about the ‘barbarous jargon of neither Welch nor English, but a bad mixture of both’ that Winifred speaks.⁹ This, as well as the author’s framing of class through dialect and education, provides a rare and useful sociohistorical picture of Bennett’s own linguistic and class biases during the early years of the popular novel form.

Ellen and Winifred in Ellen, Countess of Castle Howel

Bennett’s Ellen, Countess of Castle Howel follows the titular Ellen Meredith as she comes of age in remote northern Wales. Ellen and her maidservant, Winifred, reside with a large aristocratic family who live beyond their means in a ruinous mansion called Code Gwyn. Both of these place names adopt Welsh syntax and English spellings of Welsh words — Code for coed (‘wood’), Castle for castell, and potentially, Howel for hywel (colloquially, ‘eminent’).¹⁰ These alterations from Welsh spellings to English imply both that Bennett was specifically writing for an English-reading and -speaking audience, and that she altered the vowels, particularly with Howel and Code, in order to influence English readers’ conception of their pronunciation. Winifred Griffiths is the WelE-speaking character who functions as Ellen’s ‘faithful shadow’, childhood friend, and maidservant as Ellen marries, leaves Wales, has a child, becomes a widow, and remarries (I, p. 41). Winifred’s WelE is put to work unremittingly throughout the novel’s four volumes, as she opines on the scandals and adventures which Ellen faces throughout the narrative. It is the prevalence of Winifred’s WelE as a defining characteristic that makes the character leap from the page as Bennett’s idea of a working Welsh woman (III, pp. 155, 190).

Anna Maria Bennett

Bennett (d.1808) ‘was probably the daughter of David Evans of Merthyr Tudful [sic], Glamorgan, who has been described as a customs house officer and a grocer’.¹¹ According to the memoirs of the actor Charles Lee Lewes, Evans moved from Wales to Bristol when Bennett was a child.¹² She was later ‘briefly married to Mr Bennett, a custom house officer’.¹³ Bennett’s name appears as ‘Agnes’ in multiple print sources, including Dorothy Blakey’s The Minerva Press, where Blakey cites the Dictionary of National Biography (DNB) entry as her source.¹⁴ I use Anna Maria because, in a letter by Bennett held in the George DeForest Collection at the Louisiana State University (LSU) Special Collections, addressed to Admiral Sir Thomas Pye, the author signs her name as ‘Annamaria Bennett’.¹⁵

Bennett was quite socially mobile throughout her life, but always on the fringes of respectability. From all accounts, she was raised as the daughter of a grocer and married Mr Bennett before becoming housekeeper and mistress to Admiral Sir Thomas Pye (1708/9–1785) for fifteen to seventeen years.¹⁶ I have been unable to find any biographical information about her faith; however, Wynn Thomas highlights Bennett’s metalinguistic hostility to Welsh Methodism in the novel Anna; or, Memoirs of a Welch Heiress (1785), where a Methodist character becomes ‘the originator of all [Anna’s] misfortune’.¹⁷ This is a notable insight because Bennett had (at the very least) close Welsh connections in the eighteenth century, and would have witnessed the Methodist religious revival of Christianity in Wales which exacerbated pre-existing stereotypes of Welshness.¹⁸ Having lived her adult life outside of Wales, this hostility to Methodism might serve to create distance between both Bennett and the Welsh characters in her work and Nonconformism in ways designed to appeal to Bennett’s reading audience.

Bennett and Pye had a daughter, Harriet Pye Esten née Bennett (1761?–1865), who became a famous actress in the late 1700s after separating from her husband, James Esten in 1789.¹⁹ According to Charles Brayne, ‘Mrs Bennett had agreed to rescue her son-in-law from bankruptcy on condition that he relinquish his wife and children and live abroad.’ This capacity of financial influence suggests that with Bennett’s career as a writer, and inheritance of a house from Pye, the author was able to live in comparative financial freedom.²⁰ Esten’s acting career took her across England, Ireland, and Scotland, where it seems she was accompanied by Bennett. By all accounts, they were quite close with Bennett helping to navigate the business of her daughter’s career. This proximity may have contributed to Bennett’s use of WelE in her novel. We know that Esten acted in numerous Shakespeare plays (e.g. As You Like It, Much Ado About Nothing). This, combined with Bennett’s reference to Shakespeare in Ellen, and the quoting of Shakespeare in The Beggar Girl and Her Benefactors (1797), suggests that the author may have been influenced by the register of WelE spoken by characters in the Henry plays such as Fluellen, whose speech features are similar to Winifred’s.²¹ In the 1790s, at the time Bennett was writing Ellen, her daughter had an affair with Douglas, the 8th duke of Hamilton, in Edinburgh.²²

While there is little mention of Bennett’s education in more recent biographical studies, some mention is found in earlier sources such as the July 1804 issue of the Lady’s Monthly Museum: ‘Mrs. Bennett is descended from a respectable Welch [sic] family, of the name of Evans, who gave her a liberal and polite education.’²³ This kind of education, which provided Bennett with a means to support herself and her children, was uncommon for female children of tradesmen at the time. W. B. Stephens does not comment directly on the time and place of occupation of Bennett’s father, but gives a useful general overview of female literacy in eighteenth- and nineteenth-century Glamorganshire: ‘as late as 1871’, in ‘mining districts female literacy lagged far behind male [literacy]’. Stephens notes that ‘daughters of skilled artisans’ had ‘poor performance [compared] to that of their brothers’, and ‘daughters of unskilled agricultural labourers […] tended to have better rates [of education] than their brothers’.²⁴ Thus Bennett’s education seems to be a unique circumstance. Blakey cites the recommendations of ‘Coleridge [who] had a very high opinion of [Bennett’s] The Beggar Girl’ (1797), and said it was ‘the best novel me judice [‘in my opinion’] since Fielding’ which speaks to Bennett’s talent for prose (p. 54). Yet, as was common for popular novelists at the time, Bennett also garnered criticism. A year after her death, her writing and education received a searing indictment in an 1809 issue of the Universal Magazine. The unnamed critic writes that ‘[Bennett’s] want of education is apparent. Her style is often inelegant, and sometimes ungrammatical. She does not appear to have had the least knowledge of any language except her own.’²⁵ Still, the amount of praise Bennett’s work received from her contemporary critics seems to outweigh the negative criticism. Indeed, both Coleridge and Mary Russell Mitford (writing to Hofland) compare Bennett favourably with Jane Austen.²⁶ Further, Bennett’s obituary in The Athenaeum describes her as a novelist who ‘may be ranked with a Fielding or a Richardson’ with ‘considerable knowledge of life, and the happy art of displaying that knowledge to advantage’.²⁷ Blakey lists Bennett as one of Minerva Press’s best-selling writers in the late 1700s and early 1800s, although her name faded into obscurity as the nineteenth century wore on and consumer tastes and novel genres evolved (p. 53).

What was Bennett’s ‘own’ language? Observing ideas around national identity in Cymru-produced literature, Jane Aaron writes

that by the end of the nineteenth century, on the evidence of its women’s writing at least, Wales did have a ‘notion’ of itself as a ‘separate nation’. However, [the sense of separateness] does not appear uniformly to have been the case a century earlier [when Bennett was writing].²⁸

The letters included in the DeForest Collection at LSU are all written in English, but was this because their recipient, Admiral Pye, was English, or because Bennett’s primary language was English? Since we do not know for how long Bennett stayed in Wales as a child, or indeed, if she was even born in Wales, the question of whether Bennett grew up speaking primarily English or Welsh is speculative. In his 1879 survey of language in the British Isles, E. G. Ravenstein states that ‘Merthyr Tydvil [sic]’ is one of several towns ‘in the interior of the county [Glamorgan] [to] boast a majority of Welsh speaking inhabitants’. He also notes that, as of 1879, since ‘many English and Irish immigrants have settled in these towns, […] Welsh is unanimously said to lose ground among the younger generation’.²⁹ This implies that, in the eighteenth century, when Bennett may have lived in Merthyr Tydfil, Welsh was the predominant spoken language, despite its rapid growth as an industrial town. Summarizing the linguistic geography of Wales between 1750 and 1851, W. T. R. Pryce says there is ‘Cymru Gymraeg or Welsh Wales in the north and west; and Cymru ddi-Gymraeg or anglicized Wales in the east and south’.³⁰ Pryce plots this data by cataloguing where Welsh ‘was the sole or dominant language’, when the ‘British census was administered in Wales [although it] did not collect any information on language until 1891’ (p. 80). In this model Glamorgan sits in the ‘anglicised south’ (p. 80). Caroline Franklin describes Wales as a ‘Nowhere from a metropolitan standpoint’ in the 1770s and 1780s, with ‘no civil institutions and […] possess[ing] no capital or metropolis’.³¹ Possessed of a ‘liberal and polite’ education within the amorphous language-scape of late eighteenth-century Merthyr Tydfil, Bennett becomes a type of writer Aaron describes as one of the ‘Romantic novelists [who] were writing for predominantly English audiences and reflecting their views’, and as one of the ‘Welsh women who had turned away from their culture’ to reflect on it from outside of Wales and ‘Welshness’.³² Indeed, Bennett’s description of Code Gwyn as ‘embosomed in the brown mountains of North Wales’ invokes Romantic imagery commonly used as exoticism in popular Gothic novels (I, pp. 1–2).

As an alternative to Aaron’s argument, this imagery — alongside the constant longing of Bennett’s character Winifred to return to Code Gwyn throughout Ellen — could indicate Bennett’s yearning or nostalgia for Wales as an adult who lived primarily outside the country. Perhaps situating Winifred and Code Gwyn in North Wales is Bennett’s attempt to make the character a part of, to use Pryce’s term, Welsh Wales. As Aaron observes,

comparatively little is known, in Wales as in England, of literature published during the revolutionary years which deployed Welsh contexts. At least insofar as it involves women’s writing, this neglect may have to do with the fact that, from the contemporary, twenty-first-century point of view, these texts, paradoxically, are not Celtic enough. Few of them were written by women from Wales, and the few that were came from the pens of writers who had chosen to leave the country and language of their birth behind them, and who were struggling to make a name and living for themselves within English culture. (p. 10)

Bennett’s persistent focus on and reference to Wales throughout her novels may also convey that she was uniquely familiar with the country and its people, as well as with stereotyped English ideas about Wales and Welshness that appealed to her Minerva Press market.

Bennett writes other Welsh characters, such as Winifred’s uncle, the ‘grey-headed butler’ Edward Griffiths, as possessing non-standard speech with no WelE indicators (I, p. 2). Griffiths’s speech contains, instead, English archaisms like ‘thou’ and ‘wilt’ (I, p. 127). In contrast Winifred’s WelE is described as being neither Welsh nor English and therefore uses speech features that require consistent, unique portrayal on the page. Thus, through Winifred’s distinct WelE speech features, Bennett represents the bilinguality of the Welsh > English dialect soundscape in her remembered and imagined Wales. Winifred’s WelE examples heavily feature pre-established, widely recognizable WelE literary dialect tropes, echoing those in the Welsh Fluellen’s dialogue in Henry V (1599). Interestingly, Winifred’s WelE features do not much compare with those of Scottish author Tobias Smollett’s loquacious Welsh servant, also named Winifred, in the epistolary novel Expedition of Humphry Clinker (1771). There are, however, similarities with the WelE in Smollett’s earlier Roderick Random (1748), spoken by the character, Morgan: namely, /p/ > /b/ substitutions.

Bennett’s metalanguage throughout Ellen is a view into the author’s opinion on both the positionality of WelE as a character feature and on education for women. Ellen’s education — how ‘she had learnt to read of the Rector, to write of his Clerk, to dance of old Griffiths, to sing and play of the family Harper, and to ride of the Bailiff’ — is considered so extraordinary that, before she can marry Castle Howel, she is sent to Mrs Forrest’s boarding school for girls to correct it (I, p. 52). Indeed, according to Bennett, Lord Castle Howel ‘considered [Ellen’s] want of education as a serious misfortune, which he would be happy to redress, by instantly sending her to Bath, at his own expence, and give her every advantage which the first instructions [in England] could bestow’ (I, p. 92). In Ellen, education, according to this framework, is integral for women to traverse social spheres within and beyond Wales.

Free indirect speech

Fludernik defines free indirect speech as a ‘frame phenomenon with a variety of different linguistic elements confirming or triggering a free indirect discourse reading’. FIS is, she argues, ‘the coordination of several [syntactic] signals’ working together to create a liminal form of character speech that exists between narrative and clearly indicated dialogue.³³ One of the clearest features of FIS is point-of-view change, usually entailing a shift in point of view from narrator/narrative to character in a way that disrupts the narrative passage with character voice (Table 1). The range and sophistication of Bennett’s FIS is a remarkable addition to what is known of the historical usage of the feature. Bray discovered the existence of FIS in Bennett’s earlier epistolary work, Agnes de-Courci: A Domestic Tale (1789), which demonstrates Bennett’s skill as an author in her ability to use FIS in both epistolary and third-person narrative forms.³⁴ According to Fludernik’s summary of recent research into the history of FID in eighteenth-century novels, scholars have found that the use of both FID categories, FIS and FIT, are ‘linked to a female tradition of writing’ in the late eighteenth century.³⁵ Being such a widely read and popular author at the time similarly sets Bennett up to be exceptionally influential, perhaps even on Austen’s work, in the following decade.

The syntactic conditions of Winifred’s FIS almost always involve some form of verb in the shift between narrative to FIS, followed by WelE-indicating speech. For example, ‘Winifred sighed, and went away praying to cot, to forgive her sins’ (II, p. 30). Here, the verb ‘praying’ functions as the transition between the third-person narrative and Winifred’s point of view. But not all of Winifred’s WelE FIS use these conditions. For example, in the first appearance of Winifred’s distinctive WelE, which also demarcates the transition from narrative to FIS, the passage is simultaneously the first time Winifred speaks, the first example of WelE, and the first example of FIS in Ellen. It occurs as Winifred lifts the body of the family dog, Lion (another example of Bennett using the English version of a common Welsh name Llew, which means ‘lion’), just killed by Lord Claverton, from the arms of Ellen, and speaks. The passage reads:

[Winifred] […] desired the fair mourner to be comforted; for a very good reason, all the crying in the world would not bring Lion to life; and, for her part, she was sure the strangers were negers, for none others would hurt such a harmless creature; and, inteed, she thought Master Lewis might a come to his own home, without bringing savage negers with him. (I, pp. 17–18)

Bennett’s use and spelling of the term ‘neger’ was not unique at the time — she uses it again in 1797’s The Begger Girl and Her Benefactors. The term is clearly negative, as implied by its preceding adjective, ‘savage’.³⁶ In the above passage Bennett definitively segues from narrative to FIS with ‘all the crying in the world’, and then Winifred’s distinct WelE voice is evoked with the first example of recognizable WelE — the ‘inteed’ /d/ > /t/ substitution — accompanied by a more generic non-standard ‘might a come’ where the standard English ‘have’ is reduced to a single sound, /a/. However, both the ‘desired’ and the ‘for a very good reason’ appear also to be part of Winifred’s FIS. It could be argued that Bennett is using ‘desired’ instead of wanted in the narrative as a part of Winifred’s speech (i.e., ‘I desire you to…), but it must also be asked who is saying it’s ‘a very good reason’ — Bennett’s narrative voice or Winifred? My reading, from the context of the passage, is that while not clearly meant to be dialogue, it does invoke Winifred’s voice in a liminal space between narrative and speech and is thus included in the FIS (Figs 1, 2).

Fig. 1: FIS from Winifred Griffiths and the lone closing quotation mark indicating speech in the first edition of Ellen (I, pp. 17–18).

Fig. 2: Winifred’s FIS from the first edition of Ellen has been turned into direct speech in the second edition, signified by an opening quotation mark (Ellen, 2nd edn, 4 vols (printed for Lane, Newman at the Minerva Press, 1805), I, p. 18).

In the first volume Bennett spends considerable time on metalanguage regarding WelE. Here, an example of standard English FIS also occurs, produced by Mrs Forrest, the head of the school in Bath, to where Ellen and Winifred are sent by the Merediths and Lord Castle Howel. According to Bennett, Winifred ‘spoke very bad English, and worse Welch [sic]’ (I, p. 70). This explicit metalanguage functions in two ways. First, it raises the question of what language Winifred is speaking at any given point throughout the novel, and also implies the narrator has sufficient command or familiarity with Welsh to discern ‘worse’ Welsh from any form of standardized Welsh. Second, it also sets up Mrs Forrest’s subtle FIS in the following passage: ‘though Ellen approached as near perfection as most heroines of her age, yet she certainly had a welch accent’ which to Mrs Forrest and the other ladies ‘sounded […] uncouth’ (I, p. 143, emphasis added). Bennett’s remedy for this is Mrs Forrest ‘prudently condition[ing] for [Winifred and Ellen’s] entire separation’ (I, p. 144). Here, the author categorizes WelE as ‘barbarous’, a facet of character to be ‘got rid of’ which makes it an integral flaw of Winifred’s that the latter cannot be similarly deconditioned from speaking (I, pp. 143, 144). This positioning of Winifred’s character as ‘barbarous’ for her speech aligns her situation with Lord Claverton’s, who is ‘savage’ for his act of killing the Meredith family’s dog (I, p. 17). The commonality of both of these states of savagery and barbarity, to Bennett, lies in their otherness from English: Winifred does not sound English and is thus barbaric (and Ellen must be quarantined linguistically from her); Claverton does not necessarily look English (he has just returned from the West Indies, presumably with a tan), and behaves violently, and is thus, to the ‘barbaric’ Winifred, a ‘savage’.

WelE literary dialect: consonants

The following table is a list of examples of Winifred’s most commonly used WelE speech features (Table 2). For this dataset, I define ‘common’ as occurring more than three separate times throughout the four volumes, in similar syntactic circumstances, indicating a sound pattern in Winifred’s speech. Below is a partial list of features found in the text.

Although there are some vowel changes, Winifred’s ‘barbarous’ WelE consists mostly of consonantal change (I, p. 143). Her speech contains features already established by Thomas Sheridan and Shakespeare as belonging to representations of Welsh English in literature, such as /p/ for /b/, where words like bless become pless (I, p. 89). According to Sheridan, the Welsh are ‘constantly substituting’ what he terms ‘pure mutes’. For example,

instead of b, they use p; for g, they use k, or hard c, and for d, they employ t. For blood, they say, plut; for God, Cot; and for dear, tear. In like manner […] they substitute f in the place of v; s in the place of z.³⁷

All of which are similarly found in Winifred’s speech. Sheridan’s work was quite well established by the time of Bennett’s writing, so there is a possibility that she took spellings from him. However, Winifred’s dialect is also far wider in range than Sheridan’s examples, her dialect eclipsing even Shakespeare’s WelE in Henry V. Like both Sheridan’s and Shakespeare’s WelE registers, Winifred’s consonant changes are the most consistent feature, categorically. Some other examples of Bennett’s consonant changes include /d/ > /t/ in words like laty (lady), Cot (God), tivil (devil), tear (dear), reverent (reverend). Occasionally, this substitution is the only WelE feature in some words; more frequently, however, it is accompanied by other recognizably Welsh consonant substitutions that go alongside, such as the /g/ > /k/ change in Cot. There is little pattern to this beyond the proximity of other consonants which are altered usually in a consistent manner in other words.³⁸

Metalanguage and prescriptivism on WelE

In his 1780 General Dictionary of the English Language, Sheridan offers ‘a few words to the Inhabitants of Wales; in order to shew how easily they might get rid of their provincial dialect’ (p. 62). According to Joan Beal, ‘the aim of [Sheridan and others] was to describe and prescribe what they considered correct pronunciation’ with discussions that ‘provide us with evidence of how the language of the labouring poor was regarded’.³⁹ Bennett does something similar in the first volume of Ellen. Her metalanguage regarding the negative reality of Ellen’s described (but unvisualized) Welsh accent aligns Bennett’s own view with Sheridan’s prescriptivist linguistic bias about WelE. Metalanguage revealing Bennett’s linguistic bias appears in the perspective of Mrs Forrest, the teacher whose role is to be an authority on the subject of correct speech, and whose FIS I have described previously in this article. Let us look again at the passage. It reads that Ellen ‘approached as near perfection as most heroines of her age, yet […] certainly had a welch [sic] accent, which to the refined ears of Mrs Forrest, and her ladies, sounded a little uncouth’ (I, p. 143). Winifred, according to Bennett, is a lost cause: ‘Winifred’s [speech] was a barbarous jargon of neither Welch [sic] nor English, but a bad mixture of both, which [Mrs Forrest] plainly saw would never be got rid of, she prudently conditioned for [Ellen and Winifred’s] entire separation’ (I, pp. 143–44). In the next passage Ellen herself acknowledges her own speech ‘defect’ (specifically that she speaks English with a Welsh accent) and ‘reproaches’ herself for it. Bennett is remarking on the effect of class mobility and, tacitly, the effect of gender expectations on female education by showing Ellen, through literal mobility, as being able to get rid of her ‘provincial dialect [of WelE]’ while Winifred is unable, despite the two women sharing similar mobility in situation.⁴⁰ The finishing school Ellen attends is grooming her to be ‘Countess of Castle Howel’, which is a status she cannot hold with an ‘uncouth’ accent (I, p. 143). Winifred, on the other hand, can have no aspirations beyond continued friendship and servitude to Ellen, or potential marriage to fellow servant Joseph (who speaks in a literary dialect as well). Even though Winifred is restricted by role and class from gaining anything through standard speech adoption, Bennett says she cannot anyway as ‘clearly’ her dialect ‘would never be got rid of’ (I, p. 144). Bennett’s language positions the WelE dialect, specifically a form that includes its Welsh influence as ‘barbarous jargon’, as uncouth, lacking refinement, an imperfection, a feature which cannot be disunited from Winifred’s character. It can, however, be removed from the upper-class Ellen in a rhetorical move that highlights Bennett’s class and linguistic biases.

By situating Code Gwyn as deeply ‘embosomed in the brown mountains of North Wales’, Bennett potentially does two things. Either she sketches a class-indicative dialect hailing from this region (if she was personally familiar with northern Wales, a detail which, as of writing, is unknown), or she transposes the southern WelE speech sounds with which she is familiar onto the speech of the unfamiliar north (I, pp. 1–2). Given that Bennett gives WelE dialect to Winifred, a speaker of ‘very bad English, and worse Welch’ in remote northern Wales, this engages metalinguistically with the idea Alexander Ellis would later put forward in 1889 that speakers of Welsh in rural Wales speak English as a ‘foreign language’ rather than bilingually with regional dialects in both Welsh and English.⁴¹ Bennett’s use of ‘bad’ and ‘worse’ as adjectives for Winifred’s two spoken languages implies both that there are inherently good or, at least, standard ways of speaking English and Welsh, and that the author is aware of these. Thus Winifred’s bilinguality is fully non-standard, a dialect — of English with Welsh features or a dialect of Welsh with English features. The existence of Bennett’s Ellen, among other examples of WelE literary dialect, including T. J. L. Prichard’s Adventures and Vagaries of Twm Shon Catti and Shakespeare’s Henry V, complicates Ellis’s argument that Welsh speakers of English from the rural, inner parts of Wales speak a learned ‘book English’ and only along the English borders is ‘the [WelE] more and more dialectal’.⁴² This borderland fluency and the way it exists in the intermediate auditory and geographic landscape between Welsh and English is Ellis’s idea of WelE. Some of the English in the inner part of Wales spoken as a ‘foreign language’ is merely, according to Ellis, ‘worn-out English forms without any Welsh influence’.⁴³ The cultural linguistic heritage, and how they get to this perceived ‘worn-out’ form is addressed in Ellis’s On the Delimitation of the English and Welsh Languages, in which he writes that ‘book language’ arises from ‘foreigners [in this case, he means first-language Welsh speakers] who learn a language by book and by orthoepical instruction, [and] naturally acquire the book language, tinctured, however, essentially by their own nationality’.⁴⁴ Dialect, however, according to Ellis is ‘the language of native conversation, it is […] what we learn from our fathers and mothers, our school and college-companions, the men and women with whom we daily consort’ (p. 184). Yet, according to Bennett, Winifred and Ellen are getting their English and Welsh from the ‘school’ companions and the ‘men and women with whom [they] daily consort’.⁴⁵ Thus, linguistically and geographically, Ellis’s dialect boundaries do not account for Bennett’s example of Winifred’s rural ‘welch [sic]’-accented English speech. Penhallurick argues that the ‘Anglicization [of Wales] down the centuries was aided by events which boosted the status of English and lowered that of Welsh’.⁴⁶ John Aitchison and Harold Carter note that from the sixteenth century, ‘if the Welsh gentry wished to participate in public life’, they would do so ‘in English […]. There followed the conviction that Welsh was the language of the barbarous past, English the language of the civilised future.’⁴⁷ Bennett reflects this with the linguistic judgements Mrs Forrest passes on the speech and accents of both girls.

Conclusion

Although it is not the only early example of WelE in a novel (see Rose Cecil, tentatively attributed to a ‘Lady Mary C — ’ (1797)) or FIS (see Agnes de-Coerci (1789)), the importance of Bennett’s early contribution of WelE dialect representation lies in the form’s quantity and its metalinguistic, social frame, which reveal Bennett’s own late early modern linguistic biases as a socially mobile working woman, perceived by surrounding society to also be a Welsh woman.⁴⁸ William Labov defines sociolinguistic research as the ‘obtain[ing of] a record of overt attitudes towards language, linguistic features and linguistics stereotypes’.⁴⁹ With the voluble, pugnacious Winifred, Bennett manages to capture a recognizable ‘Welshness’ in accent and character through literary dialect and metalanguage about the character’s speech. Bennett’s use of WelE to help characterize Winifred ultimately defines the fictional servant as a product of place. The other forms of non-standard Bennett writes situate the characters who speak them within class — specifically, the more distinct the non-standard, the lower the class — although no other character speaks with such consistent non-standard as the WelE and (off-page) Welsh-speaking Winifred. FIS, throughout, is a tool Bennett uses to bring the reader closer into the auditory life of the story. Bennett’s early, abundant utilization of both FIS and WelE throughout the novel makes Ellen, Countess of Castle Howel an outstanding example of the innovation taking place in the development of rhetorical strategies in novel writing as the genre was forming in the late eighteenth century. Bennett’s presence as a female writer allegedly from a Welsh labouring-class background, making skilled use of region-specific and other dialects and FIS in the same text, indeed showing them engaging with each other, in the mid-1790s (although she was likely writing the novel in 1793) currently seems to be unprecedented. Yet while Bennett’s WelE literary dialect is among the earliest being researched, it is highly likely that there are novels even earlier which contain significant and interesting examples of WelE. According to Diedre Lynch, the phenomenon of the British novel was a product of oversaturation of the market by novelists who both leaned into established tropes and sought to create new ones. Authors of the time acknowledged this with self-referential notes about how their texts either conformed, contributed to, or ascended the oversaturation.⁵⁰ Lynch writes that ‘the genre’s fate […] was to go over the same ground again and again’ because of its relative youth, and the intense mixture of popularity and disdain it immediately provoked (p. 453). This authorial self-awareness of the state of the genre at the time underscores the necessity of examining how novels outside the canon used rhetorical tools and tropes: the same tools and tropes considered to be defined by canonical literature. Currently, however, it is difficult to conduct a large-scale examination of literature beyond the canon because of the historical unimportance of popular novels and subsequent lack of heritage efforts. The appearance of sophisticated FIS and WelE in this largely disregarded source argues for a broad examination of this and other texts like it.