9 “Norfolk People Know Best”: On the Written Representation of Accents as Performed and Perceived by ‘Insiders’ and ‘Outsiders’

The concept of ‘insider’ vs. ‘outsider’ plays a significant role in the assessment and discussion throughout my paper, which – based mainly on nineteenth-century fiction – is concerned with the representation in writing of regional and social features of language in England with special reference to accents. Writers and readers can be insiders as well as outsiders, here simply defined as members or non-members of the speech community they represent in writing or interpret in reading. From the writer’s perspective, the language of a written text can be described as ‘intrinsic’ or ‘extrinsic’ to the author (cf. Hickey (2010: 9), who states that “where the language being represented is extrinsic to the author it may well be unreliable”, characterized among other things by vacuous re-spellings). A thorny issue in the analysis of dialect representation in fiction is the significance of authenticity, which is highlighted in Jane Hodson’s pioneering publication Dialect in Film and Literature (2014), among

other things offering critical views on one-sided, detailed linguistic assessments but also on the generally impressionistic appraisals by literary reviewers of the accuracy of dialect representation, "approving of those representations which they felt to be authentic, and condemning those they felt to be inauthentic" (p. 220). While recognizing the importance of Hodson's subtle critique, the general view taken in this paper is that authenticity remains an important factor.
The paper is structured as follows: a brief introduction to nonstandard language in writing as a field of study is given, followed by some observations on the concepts of 'insider' and 'outsider'. Section 4 focuses on the general problem of representing accents by means of the orthographic system ('semi-phonetic spelling') and section 5 provides a more detailed account of the representation of regional accents in 19th century England. Section 6 summarizes and discusses the efforts by insiders and outsiders and presents four case studies of 19th century writers who can be said to represent both categories. The paper is concluded by an attempt to draw some general conclusions from the somewhat conflicting data and viewpoints accounted for. Kirk (1999: 45) distinguishes two approaches to the study of nonstandard language in literary texts: … the dialectological, which uses literary texts as evidence of the spoken language and considers the significance provided by the use of the dialect and the nonstandard within the literary work as evidence for the dialect, often historical …. The second approach is stylistic, which considers how effective or realistic of speech the language in a particular text is and considers the role and effectiveness of the dialect and nonstandard within the literary work as a whole.

Some notes on the study of nonstandard language in writing
My background may be that of a dialectologist but I recognize the importance of both the above approaches. Scholars who argue that authenticity is not a major issue in analysing dialect representation in fiction emphasize that focus should be on the function of the dialect within the text. While recognizing the importance of function, the view taken in this paper is that the importance of function does not exclude the value of authenticity. In 19th-century texts as well as their adaptations for television, for example, authenticity is generally expected with regard to physical environment, dress codes, polite behaviour etc., and language should be no exception in providing an authentic setting. Hence the paper is indeed concerned with authenticity, in comparing features found in the selected texts to authentic data, above all drawn from Wright's English Dialect Dictionary (EDD) and the Survey of English Dialects (SED). There is, admittedly, a certain amount of circularity in the case of EDD in that a great many entries are taken from literary works but these were carefully selected by its eminent editor, Joseph Wright. An important token of the value of dialect in writing as a source of information in historical linguistics is the chapter The Dialects of England since 1776, (Ihalainen 1994), which is based on data contained in works by Wright and other 19th century dialectologists as well as fictional writers.
Sweeping statements are often found in literary criticism such as "Her men and women have characteristic modes of speech. Sometimes they are easy to recognize, as, for instance, by their dialect, which, incidentally, she used well" (Pollard 1965: 254, writing about Elizabeth Gaskell). In my view, such observations are of no value, let alone completely misleading and faulty ad hoc views such as Q.D. Leavis' claim that in Dickens' representations of East Anglian accents the sounds are not represented ("…a matter of vocabulary and grammar only") (cf. Poussa 1999: 34). Opinions vary on this issue, however, as briefly discussed in the conclusion of this paper.
A distinction is usually made between 'dialect literature' and 'literary dialect', the former term referring to works composed wholly, or at least partly, in non-standard dialect, produced for a local readership, whereas the latter refers to the representation of non-standard speech (almost exclusively in the dialogue) in literature otherwise written in Standard English. The novels by Elizabeth Gaskell, who was "well aware of the need to balance authenticity and accuracy against accessibility" (Beal 2006: 534) are said to exemplify the latter category. This distinction is somewhat fuzzy, in that writers of dialect literature more often than not produce themselves in Standard English as well, and novelists like Gaskell may have 'insider' knowledge about the variety they represent. It is not completely true that "dialect is a variable dependent on the demands of fictional situation rather than on the probable behaviour of an actual speaker" (Page 1988: 59). Schneider (2002: 71f), who is exclusively concerned with written texts as data for linguistic, notably variationist, studies, provides a useful and widely quoted classification of text types according to their proximity to speech, considering category (recorded, recalled, imagined, observed, invented), reality of speech event (ranging from real to hypothetical), speaker-writer relationship, and temporal distance. According to Schneider, texts should be as close to speech as possible to be of value; in addition they must fulfil certain size requirements. Literary writing is dismissed, since it normally displays 'categorical invariant usage'. Unfortunately, there has been a shortage of substantial corpora of 19th-century fiction including non-standard dialogue, in spite of the increase in dialect literature as well as literary dialect in the 19th century, related to the rise of the novel. A close reading of, for example, Gaskell's fictional texts, however, makes it obvious that they cannot be characterized as displaying categorical invariant usage. It is my opinion that her fictional representation of spoken language could be varyingly classified as any of Schneider's categories, i.e. 'recorded', 'recalled' etc.
The focus on factual knowledge above together with the plea for authenticity may have given the reader of this text the impression that authentic representation as propagated here should be characterized by minute phonetic detail. This is by no means the case. The concept of 'enregisterment', as recently developed by Joan Beal (2009) and others, whereby specific, often somewhat 'levelled', linguistic features become associated with a particular variety and 'reified', is not in opposition to authenticity as viewed in this contribution.

'Insider' versus 'outsider' writers and readers
In his frustration after trying to include some phonetic symbols in rendering Cockney accents in Pygmalion, G.B. Shaw referred to "… this desperate attempt to represent her dialect without a phonetic alphabet" which "must be abandoned as unintelligible outside London" (Pygmalion, Act I). It would appear that Shaw believed only 'insider readers', i.e. Londoners, would be able to understand his Cockney representations. This may seem plausible enough, but the ability to read and interpret dialect in writing is more complicated and requires more insight than simply being a native speaker. The roles of 'outsiders' and 'insiders' can be discussed from a variety of perspectives, such as the phonological level of representation, language awareness and attitudes, and the ideology underpinning the wish to write and read dialect texts. This is illustrated in the following section by examples representing different time periods (occasionally beyond the 19th century for clarification of some particular issue), genres, and social/regional speech communities.
A number of distinguished writers have represented varieties of language of which they have insider knowledge as well as varieties of which they have acquired knowledge or limited experience. Elizabeth Gaskell, for example, wrote two novels (Mary Barton and North and South) set in Manchester, her hometown, but also a novel set in Whitby (Sylvia's Lovers), where she had spent only a fortnight. Another example of a writer who wrote as an insider as well as outsider is Fanny Burney, whose novel Camilla contains representations of a number of different regional accents. Charles Dickens, as we all know, included speakers of accents from various parts of Britain as well as abroad in his novels. He has often been accused of lacking linguistic insight and reliability and was, unlike Gaskell and Eliot, not accepted as an 'informant' by Joseph Wright for the English Dialect Dictionary. Such a severe assessment is somewhat unjustified, as shown by Poussa (1999) and in the sadly neglected monumental work Sound and Symbol in the Dialogue of the Works of Charles Dickens (Gerson 1967).
Even though writers producing non-standard spellings may be quite knowledgeable about the variety they want to represent, they often fail miserably due to the inadequacy of the orthographic system. Tennyson's elaborate use of 'outlandish' spellings in his Lincolnshire poems, for example, is known to have made them largely inaccessible to the general reader. This is an important issue and the limitations of the orthographic system are therefore discussed in some detail in the following section.

The problem of representing phonetics/phonology by means of the orthographic system
Whereas a near-authentic use of dialectal morphosyntax and vocabulary is -at least superficially -fairly easy, the representation of sounds, by contrast, is fraught with innumerable problems, for the writer as well as the reader. In his 1809 work on the dialect of Bedfordshire, for example, Batchelor describes "the Deficiencies of the English Alphabet, when applied in the Explanation of provincial Errors of Pronunciation" (Zettersten 1974: 157). It is an indisputable fact that the only way to truthfully represent the actual pronunciation of vowels and consonants in writing is by using a phonetic transcription (since the late 19th century preferably the International Phonetic Alphabet (IPA)), as exemplified in a simple but correct way in the title of Tony Harrison's famous poem Them and [uz]. Understandably, this is not a generally conceiva-ble alternative for publishers, writers and readers (see the Shaw quotation in the previous section).
The following examples will illustrate some aspects of the problem through various employed strategies: a) Aj häv to tjildren ('I have two children'; from a textbook in English for Swedish emigrants in the 1890's) b) <foot> vs. <strut>; <arrm> vs. <ahm> ; <the rang spee-oon>; <a stee-an hoose> (Trudgill 1990) c) laugh (laaf||laf (regional variants)); one (wun)||Northern England also won) ( In spite of their 'outlandish' appearance, all the above examples arewith a varying degree of sophistication -exponents of some knowledge of the represented accent; yet they cannot be characterized as very successful. Whereas the 'transcriptions' no doubt make perfect sense to their creators, they are bound to be misleading, if not impenetrable, at the receiver end. Although they all are meant to represent varieties of English, some -notably a), d) and e) -clearly presuppose knowledge of the creator's variety, in the case of a) even another language. In a manner of speaking they can -with the exception of a) -be said to have been produced by 'insiders' with little or no consideration of the 'outsider' audience. More often than not, however, the transcriptions tend not to be transparent to the insider audience either.
As for d), which from a British perspective is a very amusing and adequate representation of an American accent, including rhythmical features, it "would be rather impenetrable for an American audience" (Wells 1982: 529). Conversely, e) includes apt observations of characteristic realizations and non-realizations of r in Received Pronunciation, which would be below awareness of the speakers of this accent.
Examples b) and c), both designed by linguists for pedagogical reasons, certainly demonstrate more insightful attempts at 'semi-phonetic spellings ' (cf. Beal 2006: 531), which implies serious attempts at suggesting alternative pronunciations. Of a somewhat different nature is so-called 'eye-dialect', here used in the sense of 'respellings which reflect no phonetic facts', such as sez for 'says', wimmin for 'women', including representations of allegro speech such as 'cause, 'bout, showing natural phonetic processes (richly demonstrated in local glossaries, for example).
A reasonably successful system of using near-exclusively letters of the alphabet to indicate pronunciation was created by J.C. Wells for the Reader's Digest Great Illustrated Dictionary (1984) (cf. example c), where [ə] (for 'schwa', the most common vowel in unstressed position) is the only symbol taken from outside the alphabet. Some regional features, "considered standard in a particular region" are also catered for: laugh, for example, is transcribed as (laaf||laf) and one is presented as (wun||Northern England also won), in an unsuccessful attempt to indicate the 'FOOT-STRUT split' as featured in Tony Harrison's [uz], i.e. the lack of a phonemic opposition between the vowels of these words and others belonging to the same sets (Wells 1982: 350f).
Similarly, for the first edition of his introductory textbook The Dialects of England (1990), Trudgill designed a system consisting of alphabetic letters, exemplified in b), in which the FOOT-STRUT split is shown as (oo) vs (u), for example. In the second edition of his book, however, "at the request of many readers", he felt the need to complement these transcriptions with IPA versions, no doubt a justified step in the case of a linguistic textbook. Having taught dialect courses based on this textbook for a number of years, I can testify that students kept begging for phonetic transcriptions (not a very common experience). Trudgill's fairly detailed orthographic system is of special interest here, however, since he exemplifies most of his presentations of regional dialects by means of literary texts, as in the following extract from 'The Lincolnshire Poacher' (Mabel Peacock, 1890s): But I'd rather be doon wheare th'fire An' brimstun foriver bo'ns, An' just goä roond wi' a bucket An' give fook drink by to'ns -Then sit I' yon stright made heaven, Wheare saints an' aängels sing … Here the semi-phonetic spellings are the poet's own, but my general impression is that Trudgill, at least to some extent, has been inspired by genuine dialect writing in designing his own system.
Tennyson's dialect poetry, with its elaborate use of 'outlandish' spellings of the same character as 'The Lincolnshire Poacher', has been characterized as "largely inaccessible to the general reader because of an unsuccessful attempt by the poet to indicate the precise nature of the sounds of his native dialect" (Tilling 1972: 108). This is probably due to lack of phonetic insight as well as the inadequacy of the orthographic system, but above all to the poet's overenthusiastic, unrealistic attitude as a committed, 'evangelizing' insider.
Semi-phonetic spelling is also a long-standing concern of the BBC pronunciation unit (cf. http://www.phonetic-blog.blogspot.com/). In his assessment, dated 9 December, 2011, of its recommendations, Wells draws attention to some particular problems in finding satisfactory symbols in respelling systems for English, including -the PRICE vowel, for which neither y nor igh is unambiguous, while ı has a diacritic -the MOUTH vowel, for which both ou and ow are ambiguous (cf. soul, show) -the GOAT vowel, for which oh may wrongly suggest a short vowel and oa, ou, ow are ambiguous (cf. broad, loud, now) -schwa. If oh represents a long vowel, how can we make it clear that uh represents a short weak one?
In the following section the limitations of semi-phonetic spelling are further discussed in some detail in connection with an 'inventory' of the representation of accent in 19th century fiction, the main purpose of which is to provide a background to understanding the case studies and general discussion presented towards the end of the paper.

Regional accents in 19th century England -factual knowledge and fictional representation
In comparison with earlier periods, the factual knowledge of English accents and dialects as they were spoken in the nineteenth century is more than significant. The first truly ambitious as well as insightful attempt at mapping English dialect areas was that of Alexander Ellis (Ellis 1889), but of even greater importance within the context of the present paper is Joseph Wright's English Dialect Dictionary (EDD) (Wright 1898(Wright -1905, which contains more information, is more accessible (especially after its recent digitization), and is largely based on examples drawn from fiction, generally representing insiders. The fact that Wright, with his monumental knowledge of English dialects, has included a word form constitutes a guarantee of its real-life existence but there is, admittedly, a danger of circularity here.
An important factual source of another kind is the Survey of English Dialects (SED) (Orton et al. 1962-71), based on fieldwork in the mid-20th century but in its focus on old speakers reflecting regional/ nonstandard usage not too distant in time from the fictional representations featuring in this paper. Indeed, according to Ihalainen (1994: 205), "no radical changes took place in English dialects in the post-1776 period until the second half of the twentieth century". The fairly recent, widely quoted mapping of 'Traditional Dialect areas' (Trudgill 1999) largely draws on SED data and is also very relevant for this presentation. His maps confirm, among other things, that the major division is a northsouth one, e.g. demonstrating the FOOT-STRUT split as mentioned above. As for fictional representations of regional dialects, however, it should be pointed out that those referring to the south and middle of England are neither as easily found nor as well researched as those of the north. This may be due to a perception of general southern features as connected with the standard, whereas northern speech is 'marked'.
Trudgill's mapping based on phonological criteria defines a staggering number of dialects and subdialects, such as "Southern Eastern Central East". It is not the purpose here, nor would it be possible, to describe and exemplify all these varieties through fiction. Rather, from a selection of characteristic features, I will demonstrate how writers havesuccessfully as well as unsuccessfully -tried to represent different kinds of phonological features, considering the limits of semi-phonetic spellings. Interestingly, but perhaps not surprisingly, the general north-south distinctions, such as the FOOT-STRUT split, do not appear to be represented (cf. Wales 2010: 70, however). Searching for examples is not an easy task, since text corpora have, until recently, not included dialect literature and deliberately shunned literary dialect. Thanks to the launch of the Salamanca Corpus (http://salamancacorpus.usal.es/SC/index. html), searches will hopefully be more successful in the future. For the present study, most of the examples derive from the quotations found in EDD. Unfortunately, however, regional areas are very unevenly represented in the dictionary, as recently demonstrated by Praxmarer (2010).
In sociolinguistics and dialectology, vowels have generally attracted more attention than consonants. This also appears to be characteristic of literary representations; the reason may be that it is relatively easier to create semi-phonetic spellings by modifying vowel symbols than by drastically exchanging one consonant symbol for another. The following listing exemplifies fairly successful renderings of regional vowel features by 19th century writers in various genres (the words written in capital letters are key words, representing a category characterized by the same vowel and the words in bold exemplify a semi-phonetic spelling). By 'fairly successful' I mean that any reader (outsider as well as insider) conversant with English could be expected to perceive the intended sound quality represented by the semi-phonetic spelling. This must, unfortunately, be characterized as a qualified guess; I am not aware of any substantial study investigating the reading aloud of dialect texts.
Most of the following features are described as traditional dialect features in Trudgill 1999. It is symptomatic that out of the four vowels posing particular problems in respelling according to Wells (see above) three are realised as diphthongs. This is well illustrated in Tennyson's Lincolnshire poetry, with its excess of unusual letter combinations including diacritics and diaeresis (the pronunciation of vowels in a diphthong separately) (cf. Tilling 1972). Similarly, the bulk of Gerson's (1967)  A characteristic problem in the representation of consonants, i.e. rhoticity vs. non-rhoticity, was already touched upon in the previous section (cf. examples d) and e)). In 19th-century fiction, due to the relative lateness of the emergence of non-rhoticity in conjunction with the spelling convention, 'r-fulness' does not appear to be indicated. In The Mayor of Casterbridge, set in the West Country, Hardy occasionally marks the speech of the Scotsman Donald Farfrae, e.g. in warrld 'world' (presumably signalling focus), but never in the representation of his local speakers (whose speech was presumably also 'r-ful'). This may well be an 'insider effect'. The interpretation of r-fulness/r-lessness based on spelling must be made with caution, however, since the social significance of r-dropping remains unsettled even in the early twentieth century (Ihalainen 1994: 215) and it is the occasional presence of rhoticity that attracts attention (Wells 1982: 30).
With the exception of representations of allegro speech, such as  Gillian Brown's Listening to Spoken English (1977). Hodson (2014: 85) draws attention to Dickens' representation of the Artful Dodger's speech in Oliver Twist, in which there is … some attempt to indicate his intonation through punctuation, such as the exclamation marks which indicate emphasis, the question marks which indicate a rising intonation, and the dashes in the word 'com-pan-ion' which presumably indicate that each syllable of the word is sounded out in full.
Interestingly, and clearly related to the insider/outsider thrust of this paper, Hodson continues: All of this can be contrasted with the way in which Oliver's speech is reported in this passage, which is given the form of indirect speech as the narrator summarizes what Oliver said, without giving any flavour of how he said it … the dialect speech is 'other' and its peculiarities are highlighted, while the main narrative work is conducted in Standard English.

Insiders and outsiders in the perception, interpretation and representation of accents -a brief summary and four illustrative case studies
In studying the use of nonstandard varieties in English literature, it is important to remember that English spelling does not represent any existing dialect phonetically. By convention, therefore, when a writer uses normal English spellings in dialogue, for example, we infer that the pronunciation intended is the standard of the audience for which the work is written, while special deviant spellings indicate the pronunciation of a dialect that is not the audience's standard. This can lead to some rather unusual variations. For example, a writer representing an Irishman to a predominantly English audience might be inclined to use spelling to indicate Irish pronunciation, while the same writer might not do so when presenting an Irishman to a predominantly Irish audience. (Traugott & Pratt 1980: 339) In this final section of the paper the insider-outsider aspect, highlighted in the main title, takes centre stage. This aspect has already surfaced in various contexts; hence it makes sense to begin by summarizing what has emerged so far. This summary is followed by a few brief 'case studies' of certain works by 19th-century writers who have produced representations of accents both intrinsic and extrinsic to them: Fanny Burney, Elizabeth Gaskell, George Eliot, and Alfred Tennyson. By way of conclusion, various general aspects of the topic are briefly considered, such as the phonological level of representation, linguistic awareness and attitudes, and the ideology underpinning the wish to write and read dialect texts, including conscious effort-raising measures in, for example, the education system.

A brief summary
• Doubts and fears have been voiced as to the reliability of outsiders representing accents as well as being able to interpret/read representations of other accents than their own, intrinsic variety (Hickey 2010: 9; Shaw (Pygmalion)); • In an extended sense, the outsider/insider factor also applies to linguists as well as perceptive writers of fiction who do not envisage the problems facing the readers of their efforts (cf. examples a)-f) in section 2 above). If, for example, students exposed to the semi-phonetic spellings in Trudgill's The Dialects of England have a Yorkshire accent or have Swedish as their first language, they tend not to be able to read out the semi-phonetic spellings in the desired manner; hence they demand phonetic transcriptions. It has been pointed out that "any attempt at indicating accent through orthographical manipulation will only work if writer and reader share an understanding of the variety being so represented" (Hodson 2014: 92); • Whereas the concept of'semi-phonetic spellings' implies serious attempts at suggesting alternative pronunciations, the only purpose of 'eye-dialect' (respellings which reflect no phonetic facts, such as sez, wimmin, ennything) appears to be a signal to the reader that a character uses vulgar or nonstandard language. Consider, for example, the use of nonstandard spellings in Gone with the Wind for the speech of blacks while using standard spelling for whites, even though the speech of both groups is phonetically very similar. In this case, writers as well as readers might well be extrinsic to the variety. • As noted, for example, in connection with Tennyson's elaborate spellings, and also in the general analysis of regional features, representations are often inaccessible to the reader because of inability to indicate the precise sounds of an accent. This inability could be due to phonetic ignorance but above all to the sheer impossibility of representing phonetic detail in an orthographic transcription, which will suggest different realizations to outsider speakers of different accents. Representations like Tennyson's would presumably cause problems to insider readers as well, unless they were given special training.

Four case studies:
Fanny Burney: In her widely popular novel Camilla (1796) Fanny Burney included a chapter containing a lively account of a performance of Othello, in which all the parts except Iago were played by actors speaking their own local dialect. Cassio, for example, who is presented as hailing from Norfolk, says dewk for 'duke' (a representation of Yod-dropping, not quite as felicitous as Dickens' <oo> exemplified in section 3 above). Othello himself is said to be 'a true Londoner', as exemplified by wery, avay, and Desdemona's father, a West Country man from Somerset, produces zpeak, confez (the second example not adequately exemplifying 'voicing of initial fricatives'). The actress playing Desdemona is said to come from Worcestershire, but her speech is -confusinglyrepresented mainly through excessive H dropping as well as H insertion. Fanny Burney was known to have 'a good ear for dialect' which is, on the whole, apparent from her representations. She appears to be particularly successful in representing her home dialect (Norfolk), i.e. as an insider. It seems likely, however, that due to her exposure to other local accents through a wide circle of acquaintances in her London life, her awareness of the home dialect had been heightened. It is interesting to note -as pointed out by Blank (1996: 3) -that for Renaissance writers dialects appeared to have nothing to do with 'home'; Shakespeare, for example, never represented Warwickshire speech.

George Eliot:
Warwickshire speech as perceived by an insider some 300 years later, however, occasionally features in George Eliot's novels, for example Silas Marner. An experienced translator and editor, she had given much thought to the problem of representing local speech, as seen in the following quotation from a letter to Skeat, published in the Transactions of the English Dialect Society: It must be borne in mind that my inclination to be as close as I could to the rendering of dialect, both in words and spelling, was constantly checked by the artistic duty of being generally intelligible. But for that check I should have given a stronger colour to the dialogue in Adam Bede, which is modelled on the talk of North Staffordshire and the neighbouring part of Derbyshire. The spelling, being determined by my own ear alone, was necessarily a matter of anxiety, for it would be as possible to quarrel about it as about the spelling of Oriental names. The district imagined as the scene of Silas Marner is North Warwickshire; but here, and in all my other presentations of English life except Adam Bede, it has been my intention to give the general physiognomy rather than a close portraiture of the provincial speech as I have heard it in the Midland or Mercian region. It is a just demand that art should keep clear of such specialties as would make it a puzzle for the larger part of its public; still, one is not bound to respect the lazy obtuseness or snobbish ignorance of people who do not care to know more of their native tongue than the vocabulary of the drawing-room and the newspaper. (cf. Cooke 1883: 293) George Eliot, indeed, practises as she preaches, i.e. indicating 'the general physiognomy'; hence not much of interest with regard to phonetic detail is found in her work. Instead, she provides rich and consistent details representing local syntax and morphology, including allegro features (i' 'in', ha' 'have', wi' 'with'). Her representations, incidentally, are strikingly in accordance with Tolkien's use of dialectal Warwickshire/Oxfordshire forms in The Lord of the Rings, characterized by Johannesson (1994: 55) as "selective rather than wholesale". A comparison between the dialogues in Silas Marner and Adam Bede suggests clear but subtle distinctions between the two represented regional varieties. In Adam Bede, especially in Lisbeth's speech, there are examples of generalized definite article reduction (th') and, as stated by the author herself, there is generally a closer 'portraiture'. Hence Eliot as an outsider has produced a more detailed representation than Eliot as an insider.

Elizabeth Gaskell:
Elizabeth Gaskell was -like George Eliot -"well aware of the need to balance authenticity and accuracy against accessibility" (Beal 2006: 534). In contrast with Eliot, however, she was keen to include phonetic detail in her renderings of social and regional dialects, but an interesting shift in the character of these representations can be observed. As her writing and creative power of representing human speech and behaviour matured, she appears to have shifted from an essentially itembased approach to a more discourse-based, psychologically motivated one, yet without foregoing her linguistic intuition and knowledge.
Her linguistic reliability was recognized by Joseph Wright in that he used two of her novels (Mary Barton and Sylvia's Lovers), exemplifying two different dialect areas, as data for specimens in his dictionary. In Mary Barton, her first novel, which is set in Manchester, her home town for the last seventeen years, Gaskell somewhat didactically made a point of using certain tokens of Lancashire dialect vocabulary, for which she could provide explanations or etymologies in footnotes, often exemplifying the use of the same words by renowned writers such as Chaucer and Shakespeare. For the verb form getten, for example, a reference is made to Chaucer's Canterbury Tales: 'For he had geten him yet no benefice'. Such references seem to signal the author's explicit wish to raise the status of the dialect. As the story proceeds, she appears to have given up such elaborate references and they are not found in her later works. It has always been assumed that Elizabeth Gaskell drew a great deal of her knowledge of regional dialect from her husband, the Rev. William Gaskell, who was known to have lectured on Lancashire dialect. A close study of his lectures, however, reveals that his interest in dialect seemed to be of a more traditional character than what is signalled in the works of his wife. His articles deal exclusively with vocabulary, presented in item-based listings, including some wild etymologizing.
A close examination of the manuscript of her 'Whitby novel' Sylvia's Lovers reveals that she made a number of changes suggesting linguistic awareness, for example with regard to the use and form of the definite article, and for the second edition she 'corrected' the dialect, changing some Lancashire forms into East Yorkshire ones. It should be noted that she largely based her familiarity with Whitby speech on two weeks' holiday in the area and some later, more 'academic' consultations. In Mary Barton, published in 1848, as well as Sylvia's Lovers (1863) dialect is used extensively, including some detailed representation of segmental phonology, which is in accordance with SED findings (Melchers 1978: 116-18). If anything, there is more detail in the Whitby novel, i.e. the representation of the more 'extrinsic' variety.
The detailed representation in Sylvia's Lovers, her last novel but one, is not quite in accordance with her development as outlined above. In North and South (1855), set in Manchester, and -in particular -Wives and Daughters (1865), probably representing her most 'intrinsic' area, she develops fine nuances of social/regional differences in discourse, and the item-based features tend to be more generalized. Interestingly, the recent TV adaptations of Gaskell's novels demonstrate an awareness of 'enregisterment' in that regional and social differences are not explicit but hinted at in subtle ways and through few but recurrent linguistic features.

Alfred Tennyson:
Of the four writers featured in the case studies, Alfred, Lord Tennyson, is by far the one who has made the greatest effort to create a 'genuine' local dialect representation, resulting in a staggeringly complex array of outlandish vowel symbols. Consider, for example, the beginning of his well-known poem Northern Farmer, Old Style, the first of his poems written in Lincolnshire dialect: "Wheer 'asta beän saw long and meä liggin' 'ere aloän?" Alfred Tennyson (1809-92) was born in Somersby, Lincolnshire, but left the county for the south of England a good twenty years before he began writing poetry in dialect, paid only occasional visits to the area after that and had little contact with the speech he was trying to convey. He did, however, publish seven long poems in dialect, the last of them published posthumously. Tennyson is known to have taken great trouble in consulting experts in the field, including A.J. Ellis, who 'proofread' his poetry carefully and offered a great deal of criticism but also profited on his many discussions with the poet for the compilation of his monumental On Early English Pronunciation (1889). It is also worth mentioning that most of Tennyson's dialect poems were used as source material in another monumental publication, Joseph Wright's EDD (1898-1905. In his pioneering study Local Dialect and the Poet (1972), Philip Tilling, editor of the SED volumes on East Midland dialects, scrutinizes and discusses Tennyson's representation of dialect in detail as collated with the findings at the Lincolnshire localities. This penetrating and knowledgeable investigation demonstrates how a substantial number of Tennyson's representations correspond to SED data from other parts of England, at best from other parts of Lincolnshire (incidentally, a most complex area which probably explains some mistaken advice from the experts consulted) but also the North and the West Country. Tilling's general conclusion is that "the poems, though they contain much that seems to be genuine, cannot really be said to give an entirely reliable impression of the Lincolnshire dialect heard by Tennyson in his youth" (p. 107).

Concluding remarks
In concluding this attempt to discuss some aspects of the representation of English accents in 19th-century fiction, it seems justified, if not inevitable, to ask the following question: Who does it best -the outsider or the insider? The text so far has not been conclusive: on the one hand profound knowledge of a variety is required, but it can also lead to exaggerated narrowness in the representation with a frustrated readership, including insiders, as a result. "It is to be acknowledged that even texts by native or local writers, however informal, have potential problems as accurate or reliable sources of linguistic data, particularly phonological" (Wales 2010: 68). A close study of some appreciated fictional writers, in fact, reveals that they may often be more competent in representing dialects other than their native tongue. According to Hickey (2010: 9), who talks about 'scalar insiderness', the status as complete outsider nearly always goes together with a satirical approach. This is, however, hardly borne out by the data considered here. Consider also how Dickens, who skilfully represented an amazing number of different accents, as carefully documented in Gerson 1967, reacted when accused by the Spectator of using dialect as a means of mockery: I believe that virtue shows quite as well in rags and patches as she does in purple and fine linen, … even if Gargery and Boffin did not speak like gentlemen, they were gentlemen. (Gerson 1967: 371-2) From the readers' point of view, the more accurate the phonetic spelling, the more frustrating it will be to read. Most adults read word by word, not sounding words out letter by letter, so forcing adults to sound out nonstandard phonetic spellings would slow readers down, potentially irritating them, and thus distract them from the actual story. A native of Lincolnshire comments on Tennyson's elaborate use of dialect and 'semiphonetic' spelling in the following way: "ploughing through line after careful line, I found them as thick as porridge". On the same note: in his lecture Local Speech in Writing: Surely Nobody Reads It!, Stanley Ellis (1989: 20)) questions the value of elaborate representations of local accents even by 'insider star performers' such as Emily Brontë, wondering "whether Joseph really matters so much; if people find Emily so well worth while that they are prepared to read Wuthering Heights without even seeing the dialect bits".
In an important paper, Trudgill (1999b) shows how Norfolk Yod dropping is seen as a very salient feature by outsiders and thus represented in writing (Dickens, for example, writes dootiful for 'dutiful'), whereas insiders do not bother to change the spelling or, confusingly, produce spellings such as bewtiful (in fact, representing a traditional-dialect, closer and more centralized vowel, which is undergoing dedialectalization to [u:]) in another set of words such as boat, road, fool. Trudgill's conclusion is: "As usual, Norfolk people know best". This is not an unqualified truth, however, since insiders are known to have a tendency to seek to confirm their own preconceived notions and stereotypes, whereas outsiders may have 'fresh ears ' (cf. Melchers 1996: 164f). Ideally, linguistic investigations of local and social dialects should be carried out by insiders and outsiders working together, as recommended, for example, by Lesley Milroy (http://www.esrc.ac.uk/ my-esrc/grants/R000221074/read). The same approach would not be amiss in the representation of dialect in fiction.