Literary Dialect in Dickens: Research Report

Overview

An exploratory report on non-standard spellings and proposed speaker associations across eleven Dickens works. This phase extends the original six-work sample by five additional titles.

The complete inventory contains 11 works, 555 work–form types, 429 distinct spelling forms and 729 candidate form–speaker associations. The five additional works contribute 182 work-specific forms and 227 candidate associations.

Inventory of all eleven works

Our Mutual Friend

Speakers considered: Rogue Riderhood, Sloppy and Silas Wegg. The work contributes 33 candidate forms and 45 form–speaker edges.

The dialogue offers representations of London river work and poverty. Riderhood’s wot, reg’lar and arter contrast with Sloppy’s wos and wery. Silas Wegg’s performative recitation and pretentious speech, however, should not simply be assimilated to an undifferentiated “Cockney” category.

Riderhood’s p’raps and Sloppy’s wos are examples for which the proposed form–speaker association can be supported by identifiable speech contexts.

I do everything reg’lar. — Rogue Riderhood

I might be a minute arter him; I might be half a minute — Rogue Riderhood

I never heerd it called that. It’s a long word for it. Hows’ever, p’raps you call it so. — Rogue Riderhood

Read the original text at Project Gutenberg →

Barnaby Rudge

Speakers considered: Ned Dennis, Miss Miggs, John Willet and Sim Tappertit. The work contributes 19 candidate forms and 27 links.

Ned Dennis, the hangman, uses forms such as helegant, dewoted and picter. These may suggest hypercorrect initial /h/, socially stylised substitutions, and the staging of a speaker’s persona. Such spellings are varied and should be read in relation to the character’s social role and local discourse.

The electronic edition contains a transcription note indicating that such spellings as wery may reproduce Dickens’s deliberate rendering of dialect. This editorial evidence is valuable but still requires attention to the particular copy-text.

And a wery game old gentleman you are — Ned Dennis (hangman)

the helegant bits of work it has turned off — Ned Dennis (hangman)

I feel quite molloncholy to think it should ever grow old — Ned Dennis (hangman)

Read the original text at Project Gutenberg →

Hard Times

Speakers considered: Mr Sleary, Stephen Blackpool and Rachael. The work contributes 64 candidate forms and 74 links.

This addition contains one of the clearest contrasts between two kinds of representation. Sleary’s thquire, thith and thervant represent an individual articulatory characteristic explicitly described in the novel. Stephen’s awlus, mun, coom and wi’ have a different function, indexing a northern working-class speech style.

Sleary’s spellings must not be classified automatically as regional dialect. Nor should shared forms such as thee and ha’ be assumed to identify one geographical group uniquely.

Thquire! Your thervant! Thith ith a bad piethe of bithnith — Mr Sleary

Ay, Rachael, lass, awlus a muddle. — Stephen Blackpool

Let me see thee when I coom in. — Stephen Blackpool

Read the original text at Project Gutenberg →

Little Dorrit

Speakers considered: Mr Plornish, a Marshalsea turnkey, Mrs Plornish and Maggy. The work contributes 24 candidate forms and 29 links.

Character-linked evidence is distributed among several speakers: Plornish’s wery, the turnkey’s p’raps and ackney, and Maggy’s reg’lar. Mrs Plornish’s phrase “Me ope you leg well soon” is a performance of simplified English addressed to a foreigner and must be separated from habitual native-speaker dialect.

Unrestricted string matching of forms such as air would also capture ordinary standard-language nouns; context-sensitive disambiguation is essential.

Wery well then. — Mr Plornish

with his ed upside down — Mr Plornish

P’raps she’ll take a ackney coach. — Marshalsea turnkey

Read the original text at Project Gutenberg →

Sketches by Boz

Speakers considered: omnibus cads, Crofts, Samuel Wilkins and street performers. The collection contributes 42 candidate forms and 52 links.

As a collection of sketches rather than one continuous novel, this work often represents anonymous or occupationally identified voices. Spellings including werry, vich, vith, vos and niver are distributed across omnibus travel, street entertainment and other urban scenes.

Where a speaker cannot be uniquely identified, the desirable unit of annotation is not a fictitious generic “street character” but an event or speech span linked to a particular sketch, occupation and textual position.

werry much obliged to you indeed, sir. — Omnibus cad / coachmen

I wos a thinking of taking J’mima to the Eagle to-night. — Mr Samuel Wilkins

I don’t think it’s sich a wery great age, Sir. — Crofts (barber)

Read the original text at Project Gutenberg →

Methodological distinctions: what counts as literary dialect spelling?

For the purposes of this inventory, distinguish at least (1) vowel representation, (2) consonant substitution, insertion and deletion, (3) syllable reduction indicated by apostrophes, (4) regional lexis, (5) idiolectal patterns and (6) situational accommodation or language contact. Literary spelling is not a phonetic recording: it is often a social or stylistic construction by the author.

Limits of direct speech: quotation marks may also enclose narration, diaries, letters, direct thought, embedded speech by another character, or free indirect representation. A TEI/XML-type annotation retaining depth of quotation and speaker identity is desirable for reliable attribution.

Limits of frequency comparison: recorded counts are string matches over whole books. Before making claims about change over time or differences among social groups, one must control for the amount of individual dialogue, source length, publication chronology, variant spelling practices and the heterogeneous Gutenberg copy-texts.

Genre: the corpus includes ten novels and Sketches by Boz, a collection of reportage and sketches. The latter requires stratification by individual sketch and scene.

Priorities for further validation

1. Re-extract main text separately from prefaces, notes and advertisements.
2. Annotate speech spans in TEI/XML with speakers, quotation depth and discourse medium.
3. Preserve competing possible standardisations where ambiguity cannot be resolved.
4. Normalise confirmed speech tokens to a common denominator, such as per 1,000 words of verified direct speech.
5. Validate edition differences and possible transcription or OCR errors against controlled texts.

Sources and reproducibility

Project Gutenberg electronic texts were consulted; bibliographic links for all eleven works appear in the inventory above. For the five added works, candidate matching was performed case-insensitively with curly and straight apostrophes normalised.

The edge-level dataset (CSV) records work, speaker, orthographic form, proposed standard form, category, whole-text count, attribution status, source URL and notes. The “Sample-checked” filter on the network retains links for which at least some speech contexts support the proposed association.