Presentation on theme: "Literature and Corpora"— Presentation transcript:
1Literature and Corpora Corpus Linguistics Summer Institute, 30 June – 3 July 2008University of LiverpoolMichaela Mahlberg
2Today’s talk Literature and computer corpora ? Corpus stylistics and the theoretical contextExamplesWork in progress and further directions
3I am in a car park in Leeds when I tell my husband I don't want to be married to him anymore. David isn't even in the car park with me. He's at home, looking after the kids, and I have only called him to remind him that he should write a note for Molly's class teacher. The other bit just sort of slips out. This is a mistake, obviously. Even though I am, apparently, and to my immense surprise, the kind of person who tells her husband that she doesn't want to be married to him anymore, I really didn't think that I was the kind of person to say so in a car park, on a mobile phone. That particular self-assessment will now have to be revised, clearly. I can describe myself as the kind of person who doesn't forget names, for example, because I have remembered names thousands of times and forgotten them only once or twice. But for the majority of people, marriage-ending conversations happen only once, if at all. If you choose to conduct yours on a mobile phone, in a Leeds car park, then you cannot really claim that it is unrepresentative, in the same way that Lee Harvey Oswald couldn't really claim that shooting presidents wasn't like him at all. Sometimes we have to be judged by our one-offs.
4I am in a car park in Leeds when I tell my husband I don't want to be married to him anymore. David isn't even in the car park with me. He's at home, looking after the kids, and I have only called him to remind him that he should write a note for Molly's class teacher. The other bit just sort of slips out. This is a mistake, obviously. Even though I am, apparently, and to my immense surprise, the kind of person who tells her husband that she doesn't want to be married to him anymore, I really didn't think that I was the kind of person to say so in a car park, on a mobile phone. That particular self-assessment will now have to be revised, clearly. I can describe myself as the kind of person who doesn't forget names, for example, because I have remembered names thousands of times and forgotten them only once or twice. But for the majority of people, marriage-ending conversations happen only once, if at all. If you choose to conduct yours on a mobile phone, in a Leeds car park, then you cannot really claim that it is unrepresentative, in the same way that Lee Harvey Oswald couldn't really claim that shooting presidents wasn't like him at all. Sometimes we have to be judged by our one-offs.
5Corpus approaches to literature ? Corpus Linguistics 2005Oxford Workshop 2006Pala 2006Corpus Linguistics 2007Corpus style mailing list
6Literature in Corpora ? naturalness, mainstream, repeated patterns in a general corpus a literary text will disappear below the waves (Sinclair 2007)a novel as a unit of meaning (world of the text)text length and balancecopyrightMore by the same publisher (cf. BoE), levelling-out effect, with all these issues better stay away from literature
7Literature and computers “the gap is still immense between what readerscan do effortlessly, and what a computer can do.Scholars interested in calling on a computer toaid their research are limited to a very narrowrange of possible operations, and suchoperations still fall largely outside the mainstreamwork of literary scholarship.” (Miall 1996: online)
8Literature and language descriptions / grammars “Literature is a prime example of language in use; no systematic apparatus can claim to describe language if it does not embrace the literature also; and not as a freakish development, but as a natural specialization of categories which are required in other parts of the descriptive system. Further, the literature must be describable in terms which accord with the priorities of literary critics.” (Sinclair 2004: 51)
9Literariness?“a cline of literariness in language use with some uses of language being marked as more literary than others in certain domains and for certain judges within that domain”(Carter 2004: 69)
10Corpus stylistics: corpus linguistics + literary stylistics ? criticism of literary stylisticscorpus-based vs. corpus-driven:stylistics checklist (nouns, verbs, simple sentences,cohesion, etc.) or challenging linguisticcategories (lexical items, patterns)Challenge the categories
11Corpus stylistics: corpus linguistics + literary stylistics ? style is “distinctive: in essence, the set or sum of linguistic features that seem to be characteristic: whether of register, genre, or period, etc” ( Wales 2001: 371)can be interpreted in CL terms: corpus work is based on comparison
12Corpus stylistics: corpus linguistics + literary stylistics ? the description of deviations from linguistic norms (Leech 1985)can be interpreted in CL terms:primary deviation: norms of the language as a wholesecondary deviation: norms of literary composition, e.g. author, genretertiary deviation: internal, norms of a text
13characterising meanings of words word out oftextual contextKWICword ina specific text?
14Local textual functions local: apply to (a group of) lexical items in a (group of) textstextual: focus on lexical items in relation to features of texts
15Clusters as pointers to local textual functions Repetition as evidence of functional relevance.A cluster is a sequence of words that are used repeatedly (at the same time, on the one hand .. on the other hand, at the end of the ….)
16Clusters: counting and comparing (norms and deviations) at the end of the is one of the most frequent 5-wordclusters Englishit appeared to me that occurs 7x in GreatExpectations
17Clusters: counting and comparing (norms and deviations) General corporaDickens corpus: ~ 4.5 million words, 23 texts19th century novel corpus (19C): ~ 4.5 million words, 29 texts, 18 authorskey clusters
18Striking and long clusters in Dickens: 51 8-word clusters (min. 5) THE ANGLO-BENGALEE DISINTERESTED LOAN AND LIFE ASSURANCE COMPANY (8 words, MC)NOT TO PUT TOO FINE A POINT UPON IT (9 words, BH)(THE) UNITED METROPOLITAN IMPROVED HOT MUFFIN AND CRUMPET BAKING AND PUNCTUAL DEVLIVERY COMPANY (13 words, NN)
19corpus-driven categories for the 66 key clusters (dynamic groups, ad hoc labels) Speech 14As if 6Body parts 9Time and place 5Rest 12
20Labels D 19C THE FATHER OF THE MARSHALSEA 45 THE PERSON OF THE HOUSE THE PERSON OF THE HOUSE37THE LADY OF THE CARAVAN22MAN OF THE NAME OF1THE OLD MAN WITH A21CAPTAIN GILLS SAID MR TOOTS20MY DEAR SAID THE JEW19MR PICKWICK AND HIS FRIENDSGENTLEMAN IN THE WHITE WAISTCOAT18THE GENTLEMAN IN THE WHITE17HOW NOT TO DO IT16
21Speech D 19C DO ME THE FAVOUR TO 31 WHAT DO YOU MEAN BY 73 15 WHAT DO YOU MEAN BY7315BEG YOUR PARDON SIR SAID25UPON MY WORD AND HONOUR1I BEG YOUR PARDON SIR5611HOW DO YOU FIND YOURSELF23HOW DO YOU DO MR292YOU BE SO GOOD AS19WHAT I AM GOING TO3AM GLAD TO SEE YOU24Speech clusters contain pronouns or possessives
22Body part clusters D 19C HIS HANDS IN HIS POCKETS 90 13 WITH HIS HANDS IN HIS6012HANDS IN HIS POCKETS AND405WITH HIS HAND TO HIS312LAYING HIS HAND UPON HIS221HIS HEAD AS IF HE18THE PALMS OF HIS HANDS17HIS HEAD ON ONE SIDE304HIS HAND AS IF HE15Dickens shows preferences for clusters
23As if D 19C AS IF HE WOULD HAVE 41 2 AS IF HE WERE A 45 7 AS IF HE WERE GOING323IF HE WERE GOING TO26AS IF IT WERE A7223Speech clusters contain pronouns or possessives
24Time and place D 19C ON THE TOP OF HIS 21 AT THE UPPER END OF 23 1 AT THE UPPER END OF231ON THE OPPOSITE SIDE OF5415AFTER A GREAT DEAL OF16THE OPPOSITE SIDE OF THE7026Speech clusters contain pronouns or possessives
25IN THE COURSE OF THEA QUARTER OF AN HOURAT THE BOTTOM OF THEIN THE MIDDLE OF THEAT THE TOP OF THEON THE OTHER SIDE OFAT THE END OF THETHE OTHER SIDE OF THEUP AND DOWN THE ROOM
26What do (key) clusters tell us? Dickens uses more 5 word-clusters.Dickens seems to like certain clusters..Why?
27Local textual functions: body part clusters ContextualisingHighlighting
28Details of local textual functions Contextualising'You see, Mr Richard,' said Brass, thrusting his hands in his pockets, and rocking himself to and fro on his stool, 'the fact is, … (Old Curiosity Shop)'Let me see then,' resumed Mr Boffin, with his hand to his chin. 'It was Secretary that you named; wasn't it?'(Our Mutual Friend)"Eh? What do you say I have got of my own?" asked Mr. Smallweed with his hand to his ear.(Bleak House)
29HighligthingHe was the meekest of his sex, the mildest of little men. He sidled in andout of a room, to take up the less space. He walked as softly as the Ghostin Hamlet, and more slowly. He carried his head on one side, partly inmodest depreciation of himself, partly in modest propitiation ofeverybody else. It is nothing to say that he hadn't a word to throw at adog. He couldn't have thrown a word at a mad dog. He might have offeredhim one gently, or half a one, or a fragment of one; for he spoke as slowlyas he walked; but he wouldn't have been rude to him, and he couldn'thave been quick with him, for any earthly consideration. Mr. Chillip,looking mildly at my aunt with his head on one side, and making her alittle bow, said, in allusion to the jewellers' cotton, as he softlytouched his left ear: 'Some local irritation, ma'am?‘(David Copperfield)
30'That little man of a doctor, with his head on one side,' said my aunt, 'Jellips, or whatever his namewas, what was he about?
31CharacterClusters associated with the characterMr SnagsbyNOT TO PUT TOO FINE, PUT TOO FINE A POINT,TO PUT TOO FINE A, FINE A POINT UPON IT,TOO FINE A POINT UPONMr GuppyMAN OF THE NAME OF, OF THE NAME OFGUPPY, YOUNG MAN OF THE NAME, THEYOUNG MAN OF THECIRCUMSTANCES OVER WHICH I HAVE, OVERWHICH I HAVE NO, WHICH I HAVE NO CONTROLYOUR LADYSHIP SAYS MR GUPPYMr BagnetOLD GIRL SAYS MR BAGNET, BUT I NEVEROWN TO, I NEVER OWN TO IT, NEVER OWN TOIT BEFORE, BEFORE HER DISCIPLINE MUST BE,HER DISCIPLINE MUST BE MAINTAINED, ITBEFORE HER DISCIPLINE MUST, OWN TO ITBEFORE HER, THE OLD GIRL SAYS MR, TO ITBEFORE HER DISCIPLINE
32Mr JellybyHIS HEAD AGAINST THE WALL, WITH HIS HEAD AGAINST THEMr BucketSIR LEICESTER DEDLOCK BARONET IBY SIR LEICESTER DEDLOCK BARONETNOW SIR LEICESTER DEDLOCK BARONETINSPECTOR BUCKET OF THE DETECTIVEMr GeorgeYOUR FRIEND IN THE CITYI ASK YOUR PARDON SIRRichardAS WELL AS ANYTHING ELSE, DO AS WELL AS ANYTHING ELSEEstherI THOUGHT IT BEST TOWHEN WE CAME TO THE
33Mr VholesIN THE VALE OF TAUNTONMr JarndyceHAVE SOMETHING TO SAY ABOUT, SOMETHING TO SAY ABOUT IT, WILL HAVE SOMETHING TO SAYMiss FliteI EXPECT A JUDGMENT SHORTLYCharleyIF YOU PLEASE MISS SAIDChadbandIN A SPIRIT OF LOVERIGHT THAT I SHOULD BEYOU ARE TO US AGrandfather SmallweedMY FRIEND IN THE CITYTO LOOK AFTER THE PROPERTYKrookMY NOBLE AND LEARNED BROTHERJoWOS WERY GOOD TO ME
34And she rings for Mercury to show the young man of the name of Guppy out. But in that house, in that same moment, there happens to be an old man of the name of Tulkinghorn. And that old man, coming with his quiet footstep to the library, has his hand at that moment on the handle of the door--comes in--and comes face to face with the young man as he is leaving the room.(Bleak House, Chapter 33)
35Example of Bleak House Labels most frequent (59 out of 97) Speech labels (e.g. Esther)Character pairs and relationships (Guppy, Bagnet, Bucket, Charley, …)Body part cluster labels (Mr Jellyby)groups of characters that get labels (cf. Tulkinghorn, Lady Dedlock)point of view (Snagsby)
36Local textual functions: Clusters not automatically associated with the same functions in all texts. 5-word clusters useful for comparison: Dickens uses more clusters and cluster functions, in 19C clusters max character-cluster link in 19C = 3)the 5 functional groups tend to cover 5-word clustersdifferences between novels (BH ~ 350,000: word clusters, 59 labels, 2 As If, GE ~180,000: word clusters, 4 As If)the longer the cluster, the more text-specific, long clusters useful for literary analysis (the young man of the name of Guppy)
37general Dickens 19C novel NOTE: in examples already clusters that are not key clusters, because it is the functional groups that are relevantnovel
38Classification of cluster types into 5 functional groups, figures per 100,000 words (For each text all 5-word clusters >4 are classified =716 different types)
39Clusters and characterisation small numbers have to be treated with caution, clusters can not provide full picture of individual character, clusters only point to what is explicitly there, 5-clusters as formal starting-pointcharacterisation is a process, forming impressions of characters in our minds (Culpeper 2001)but: features of characters have to be seen with regard to textual world, behaviour interpreted against norms of the text world
40Speech clusters: definition Signals of interaction, formal criteria for definition(inferences about personality? )i have no doubt ofi do not know whati am not at allyou need not be afraid
41Speech clusters: definition Speech clusters are not the same as ‘imitation of spoken language’not a bit of it all 21 FDS/DS1 are reconciled then?' said Perker. 'Not a bit of it,' answered Wardle; 'she2 ping her hands and shaking her head. 'Not a bit of it.' 'At least, his name3 Dick. 'No I haven't,' she returned, 'not a bit of it. Don't you mind about4 ing her hands, and shaking her head. 'Not a bit of it.' 'Handford then,' suof course it's not. Is it in Africa? Not a bit of it. Is it in America? YOUon't believe it,' cried the gentleman, 'not a bit of it. It's an excuse not t
42all that sort of thing tendency for FDS/DS but: Mr. Pickwick's upright and honourable bearing, coupled withthat force and energy of speech which so eminently distinguishedhim, would have carried conviction to any reasonable mind; but,unfortunately, at that particular moment, the mind of Mr. PeterMagnus was in anything but reasonable order. Consequently,instead of receiving Mr. Pickwick's explanation as he ought tohave done, he forthwith proceeded to work himself into a red-hot, scorching, consuming passion, and to talk about what wasdue to his own feelings, and all that sort of thing; adding force tohis declamation by striding to and fro, and pulling his hair--amusements which he would vary occasionally, by shaking hisfist in Mr. Pickwick's philanthropic countenance.
43But I never own to it before her. Discipline must be maintained. Speech clusters are not the same as the ‘speech tics’, ‘character tags’, or idiolects described in Dickens criticismI‘ll tell you wot it all 6 in PPbut I never own to all 6 in BHBut I never own to it before her. Discipline must be maintained.(character tags and flat characters: same behaviour in different situations)
44politeness promoting social harmony (cf politeness promoting social harmony (cf. Leech 1983, Brown & Levinson 1987)am delighted to see youam glad to see youam much obliged to youbeg your pardon sirhow do you do mrI wish you good night
45impoliteness: attacking face (cf. Culpeper 1996) what have you got to ...Now, it so happened that Mr. Fang was at that moment perusing aleading article in a newspaper of the morning, ... He was out oftemper; and he looked up with an angry scowl.'Who are you?' said Mr. Fang.The old gentleman pointed, with some surprise, to his card.'Officer!' said Mr. Fang, tossing the card contemptuously awaywith the newspaper. 'Who is this fellow?''My name, sir,' said the old gentleman, speaking like agentleman, 'my name, sir, is Brownlow.
46'Officer!' said Mr. Fang, throwing the paper on one side, 'what's this fellow charged with?''He's not charged at all, your worship,' replied the officer. 'Heappears against this boy, your worship.'His worship knew this perfectly well; but it was a good annoyance,and a safe one.....'Now,' said Fang, 'what's the charge against this boy? What haveyou got to say, sir?’...'Swear the man,' growled Mr. Fang. with a very ill grace. 'Now,man, what have you got to say?'
47what have you got to (21) what have you got to say for yourself .. what have you got to say to me ...power relationships of characters (not only magistrate)categories of characters
48pointers to impoliteness and conflict DCorp Cwhat do you mean bywhat do you want here(cf. also key clusters)
49(OMF) The respected Mr Lammle was a bully, by nature and by usual practice. Perceiving, as Fledgeby's affrontscumulated, that conciliation by no means answered thepurpose here, he now directed a scowling look intoFledgeby's small eyes for the effect of the oppositetreatment. Satisfied by what he saw there, he burst into aviolent passion and struck his hand upon the table,making the china ring and dance.'You are a very offensive fellow, sir,' cried Mr Lammle,rising.'You are a highly offensive scoundrel. What do you meanby this behaviour?'[…]'I say,' repeated Fledgeby, with laborious explanatorypoliteness, 'I beg your pardon.’(OMF)
50'Now, sir,' said Mr Dorrit, turning round upon him and seizing him by the collar when they were safely alone.'What do you mean by this?'The amazement and horror depicted in the unfortunateJohn's face--for he had rather expected to be embracednext--were of that powerfully expressive nature that MrDorrit withdrew his hand and merely glared at him.'How dare you do this?' said Mr Dorrit. 'How do youpresume to come here? How dare you insult me?'....'I humbly beg your pardon, sir. ...context shows function
51... Mrs Quilp, to whom, after contemplating her for some time in silence, he communicated a violentstart by suddenly yelling out--'Halloa!''Oh, Quilp!' cried his poor little wife, looking up. 'How you frightened me!''I meant to, you jade,' returned the dwarf. 'What do you want here? ...
52aggression as personality trait, villains, hostile relationships but also local context: example of PPpositive impoliteness, abuse, paraphrases
53… Dickens very conveniently was a prolific writer…
54Joint work with Dan McIntyre 5 Bonds a bit more than Bleak House, Casino less than 50,000
55Aims of the Bond Project Why Bond?popular fictionunder-represented within stylistics (though see Ryder 1999; Montoro 2007)Some literary critical interest (Lindner 2003, Comentale et al. 2005) though considerable lack of focus on the textWhy focus on Casino Royale?First book in the series; film version released this year, bestsellerWhat questions do we want to investigate?What are the stylistic characteristics of Casino Royale?How is the character of Bond constructed linguistically?Does a corpus stylistic analysis tie in with literary critical comment on Bond?P
56Key semantic domains via Wmatrix assigns semantic tags, 21 discourse fieldsCorpus based and corpus driven, groupings for key words,
57Top 10 key semantic domains Casino Royale compared against BNC Written SamplerSemantic domainLog-likelihood valueExamples(1) Pronounsit, he, his, him(2) Anatomy and physiologybody, arm, profile, chin(3) Games391.55casino, gambler, croupier(4) Light348.72light, illuminated, sunshine(5) Unmatched217.84salle, privee, caisse(6) Furniture and household fittings171.18table, stool, pillow, bed(7) Degree146.83as [high as], as [serious as](8) Parts of buildings137.51vault, passages, doors(9) Location and direction131.71away from, top, left(10) Darkness104.00dark, unlit, darkness
58References to Bond’s BODY Given the precoccupation with Bond’s body in the book, it’s perhaps not surprising to find this translated into the latest screen version!!!Complemented with key word analysis… Key words more what Mike…
59Key words: building blocks for the world of the text fewer key words and more selective than Wmatrix183 words key wordsin category to start analysis‘Anatomy and physiology’groups of key words built bottom-up (by means ofconcordancing)
61physical aspects of the casino (table, cards, plaques, shoe, rail) people and their roles in the casino (players, croupier, spectators)aspects of games (maximums, bet, slipped)numbers and money (neuf, notes)French words and phrases in the gamesMessieurs, mesdames, les jeux sont faits. Un banco de cinq cent mille.KWs need to be looked at in context, as they can have various meaning and the most obvious may not be the one for the book.
62Spying gunmen, gunman, organization, smoke-bomb, Spying gunmen, gunman, organization, smoke-bomb, double, bureau, agent, memorandum, 007, MWDtrace… at Royale trying to trace the Jamaican millionaire …… it would take hours to trace the ownership to him… There was no trace of the gunman, …… Next he examined a faint trace of talcum powder on the inner rim … effort to find out, mystery… with a trace of impatience …… Le Chiffre showed no trace of emotion. … characterisationsome of them could also be grouped as namestrace has several meanings, and may be key here because of the occurrences of [example]both VB and Nsecond set more ‘novel specific’ in general
63concierge, patron, sommelier places boulevard, night-club, Settings and propspeople/jobsconcierge, patron, sommelierplacesboulevard, night-club,coast, villa, side-roadfood, drink,champagne, caviarclothespyjama-coat, dinner-jacketfurniturearm-chairchairVesper and Bond having a mealLe Chiffre’s benzedrine inhalercane chair and cane carpet-beater in torture scenesettings and props provide context for overall action, possible to attempt grouping according to semantic fieldsis this similar to Wmatrix only by hand? So is this useful ?Wmatrix ‘carpet-beater’ - furniture and household fittingsin the novel but can also focus on specific scenes, or items that appear less easy to classify
64Themes and characterisation luck A1.4 Chance, luckblack O4.3 Colour and colour patternsvillains G2.1- Crimeevil G2.2- unethicalall villains (7) and most evil (15 of 19) in chapter 20Themes and characterisation:Part of the meanings ‘conventional metaphorical’ -> black, selection in novel
652) Bond and Vesper/women Key word luck1) Bond as a gambler… Bond had always been a gambler. ..., and he accepted the fact, he would be brought to his knees by love or by luck…. (Chapter 7)2) Bond and Vesper/women“Perhaps I will bring you luck” (Chapter 5)… he felt vague disquiet. On an impulse he touched wood….(Chapter 5)luck not simply in category ‘games’
66Key words black, villains, evil The villains and heroes get all mixed up….. Now in order to tell the difference between good and evil, we have manufactured two images representing the extremes – representing the deepest black and the purest white – and we call them God and the Devil (chapter 20) Clothes Vesper: black dress, black velvet skirt, black velvet ribbon on her hat, black hair Bond: black satin tie, black hairnot surprisingly new meanings, but the way they link in with each other in the text contributes to the creation of the world of the textonce you begin looking at these key words, it seems a very strong/obvious picture, velvet = deceivingly nice, smooth, Vesper as traitor and dangerkey scence, wants to resign
67in KW list Anatomy and physiology category not so clearly visible, but: 'Well, when you get back to London you will find there are other Le Chiffres seeking to destroy you and your friends and your country. […] And now that you have seen a really evil man, you will know how evil they can be and you will go after them to destroy them in order to protect yourself and the people you love. […] You may want to be certain that the target really is black, but there are plenty of really black targets around. There's still plenty for you to do. And you'll do it. And when you fall in love and have a mistress or a wife and children to look after, it will seem all the easier.' Mathis opened the door and stopped on the threshold. 'Surround yourself with human beings, my dear James. They are easier to fight for than principles.’ He laughed. 'But don't let me down and become human yourself. We would lose such a wonderful machine.’ (Chapter 20)also linked through black clotheskey scence (cf. luck scene)
68Conclusionscorpus findings can provide insights into the ‘world of a novel’the need for local descriptionslinking linguistic and literary discussion
69Future work: bottom-up and top-down: reception effects on reader script and film
70Further detail….Mahlberg, M "Corpora and translation studies: textual functions of lexis in Bleak House and in a translation of the novel into German". In V. Intonti, G. Todisco, and M. Gatto (eds), La Traduzione. Lo Stato dell'Arte. Translation. The State of the Art. Ravenna: Longo,“Clusters, key clusters and local textual functions in Dickens”, Corpora.“Corpus stylistics: bridging the gap between linguistic and literary studies” In Hoey, M., Mahlberg, M., Stubbs, M., Teubert, W. forthcoming. Text, Discourse and Corpora. Theory and Analysis. London: Continuum.-- forthcoming, 2007 “A corpus stylistic perspective on Dickens’ Great Expectations”. In M. Lambrou and P. Stockwell. Contemporary Stylistics. London: Continuum.