literary lab pamphlet 1

Upload: ze-antonio-oliveira-salome

Post on 04-Apr-2018

223 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/30/2019 Literary Lab Pamphlet 1

    1/2901

    01

    LiteraryLab

    Pamphle

    SarahAllisonRyanHeuser

    MathewJockersFrancoMorettiMichael

    Witmore

    Quaniaive Formalism:

    an Experimen

    1

    AB

    January 15, 2011

  • 7/30/2019 Literary Lab Pamphlet 1

    2/29

    02

    Pamphles of he SanfordLiteraryLab

    IISSN 2164-1757 (online version)

    IISSN 2164-3431 (prin version)

  • 7/30/2019 Literary Lab Pamphlet 1

    3/29

    1

    Sarah Allison

    Ryan Heuser

    Mathew Jockers

    Franco Moreti

    Michael Wimore

    Quaniaive Formalism: an Experimen

    This paper is he repor o a sudy conduced by ve people our a Sanord, and one

    a he Universiy o Wisconsin which ried o esablish wheher compuer-generaed

    algorihms could recognize lierary genres. You ake David Coppereld, run i hrough a

    program wihou any human inpu unsupervised, as he expression goes and ... can

    he program gure ou wheher is a gohic novel or a Bildungsroman? The answer is,

    undamenally, Yes: bu a Yes wih so many complicaions ha i is necessary o look a he

    enire process o our sudy. These are new mehods we are using, and wih new mehods

    he process is almos as imporan as he resuls.

    1. Prologue: Docuscope Reads Shakespeare

    During he Fall o 2008, Franco Moreti was visiing Madison, where Michael Wimore in-

    roduced him o work he and Jonahan Hope had been doing on Shakespeares dramaic

    genres, using a ex agging device known as Docuscope, a hand-curaed corpus o sev-

    eral million English words (and srings o words) ha had been sored ino grammaical,

    semanic and rheorical caegories.1

    1 See Jonahan Hope and Michael Wimore, The Very Large Texual Objec: A Prosheic Reading o Shakespeare,

    Early Modern Literary Studies9.3 (January, 2004): 6.1-36; Wimore and Hope, Shakespeare by he Numbers: On he

    Linguisic Texure o he Lae Plays in Early Modern Tragicomedy, eds. Subha Mukherji and Raphael Lyne (London:Boydell and Brewer, 2007), 133-53; Hope and Wimore, The Hundredh Psalm o he Tune o Green Sleeves:

    Digial Approaches Shakespeares Language o Genre, Shakespeare Quarterly61.3, Special Issue: New Media

    Approaches o Shakespeare, ed. Kaherine Rowe (Fall 2010): 357-90; and Wimores blog, www.winedarksea.org.

  • 7/30/2019 Literary Lab Pamphlet 1

    4/29

    2

    Docuscope is essenially a smar dicionary: i consiss o a lis o over 200 million pos-

    sible srings o English, each assigned o one o 101 uncional linguisic caegories called

    Language Acion Types (LATs).2 When Docuscope reads a ex, i does so by looking

    or words, and srings o words, ha i can recognize ha is o say, ha i can mach o

    one o is 101 LATs. When his happens, he associaed LAT is credied wih one appear-

    ance. For example, since Docuscope assigns I and me o he LAT FirsPerson, heiroccurrence in a ex is recorded as an appearance o he LAT FirsPerson.3

    Based on hese couns, Hope and Wimore used unsupervised acor analysis a acor,

    here, being a patern ha includes some caegories, in variable proporions, and excludes

    ohers o creae porrais o received genre disincions such as hose made by he edi-

    ors o he Firs Folio (Heminges and Condell), and o he genre o lae romances ha was

    rs idenied in he nineeenh cenury. Mulivariae analyses and clusering echniques

    made groupings o he plays ha corresponded no only o convenional genre group-

    ings, bu also picked ou exs ha criics had idenied as ouliers. 4 Thus, in clusering

    Shakespeares Folio plays, he program managed o ake Henry VIIIou o he Hisory playscluser and place i near oher lae plays, a re-adjusmen rom he iniial Folio designa-

    ions ha laer criics have advocaed as well. One can see his grouping patern in gure

    1 below, aken rom an early complee linkage clusering o he plays.

    Aer seeing hese resuls, Moreti asked Wimore wheher he would consider clusering nov-

    elisic genres. Wimore agreed, and a meeing was planned or February 2009 a Sanord.

    2 For Docuscope, see David Kauer, Suguru Ishizaki, Brian Buler, Je Collins, The Power o Words: Unveiling the

    Speaker and Writers Hidden Cra (Lawrence Erlbaum Associaes: New Jersey and London, 2004). A ascinaing

    discussion o how he program came o be designed and an early prcis o is caegories can be ound a: htp://

    www.beterwriing.ne/projecs/ed01/dsc_ed01.hml, accessed 3 March 2010.

    3 Because o he way hey are used in he program, LATs mus be given names wihou spaces. Obviously he char-

    acerizaion o he words ha are conained in each o hese caegories is a mater o inerpreaion, as is he choice

    o hose words hemselves, which ook place over he course o almos a decade o hand-coding. In general, Wi-

    more and Hope use he caegories or LATs o ideniy saisical paterns, hen move rom he caegories o concree

    exual insances in order o see how paricular words are uncioning in conex.

    4 They discovered, or insance, ha Shakespeares lae romances were disinguished, linguisically, rom hose

    ha wen beore hem by word paterns ha allowed speakers o narrae pas acion while highlighing heir own

    emoional sance wih respec o hose acions (a process hey called ocalized rerospecion). Specic linguisic

    eaures o hese plays were responsible or his eec, or example (1) cerain ypes o subordinaed conjuncion

    (a comma, ollowed by he word which) and (2) pas ense verb orms inroduced by a pas ense auxiliary orm o

    he verb o be. Comedies and hisories were also shown o be signicanly disinc rom one anoher, wih comedy

    possessing a high degree o rs and second person pronouns (classed under he LATs FirsPerson and DirecAd-

    dress), a high degree o language expressing uncerainy (he LAT Uncerainy); an absence o nouns and verbs

    used o reer o moion, he properies o sensed objecs, and sensed changes in objecs (LATs labeled Moions,

    SensePropery, SenseObjec); an absence o rs person plural pronouns (he LAT Inclusive); and an absence owords indicaing social eniies or expecaions ha mus be shared or muually acknowledged (he LAT Common-

    Auhoriy).

  • 7/30/2019 Literary Lab Pamphlet 1

    5/29

    3

    3 12 9 13 6 32 29 14 2 7 25 26 2 128 3 6 31 3 4 2 10 8 1 33 4 7 11 5 15 17 20 1 6 22 23 18 1 9 24 3035

    Cluster Analysis of Folio Plays

    Observations

    A Midsummer Nights Dream (3)

    Twelh Night (12)

    Much Ado About Nothing (9)

    Two Gentlemen (13)

    Measure for Measure (6)

    Othello (32)

    Julius Caesar (29)

    The Winters Tale (14)

    Cymbeline (27)

    Antony and Cleopatra (25)

    Coriolanus (26)

    Henry VIII (21)

    Hamlet (28)

    Troilus and Cressida (36)

    Macbeth (31)

    Timon of Athens (34)

    Alls Well That Ends Well (2)

    Taming of the Shrew (10)

    Merry Wives of Windsor (8)

    A Midsummer Nights Dream (1)

    Romeo and Juliet (33)

    Comedy of Errors (4)

    Merchant of Venice (7)

    The Tempest (11)

    Loves Labours Lost (5)

    1 Henry IV (15)

    2 Henry IV (17)

    Henry V (20)

    1 Henry VI (16)

    King John (22)

    Richard II (23)

    2 Henry VI (18)

    2 Henry VI (19)

    Richard III (24)

    King Lear (30)

    Titus Andronicus (35)

    Figure 1: Dendrogram illusraing clusering o Shakespeare plays raed on Docuscopes Language Acion Types

    (LATs) produced in 2003. Clusering mehod: complee linkage, Euclidean disances. Noice he presence o

    comedies in he rs and hird columns, lae plays and ragedies in he second, and hisories in he ourh and

    h. Incorrec classicaions such as Othelloand Loves Labours Lost are discussed on Wimores blog, www.

    winedarksea.org.

    2. February 2009: Docuscope Recognizes Novelisic Genres

    The saring poin o he sudy was a corpus o 250 19h cenury Briish novels rom he

    Chadwick-Healey collecion.5 Working wih exising genre bibliographies, Moreti pu

    ogeher a sample o 36 exs loosely comparable o he Shakespeare corpus o he rs

    Docuscope experimen, which comprised 12 genre ses, divided ino wo groups o 6. The

    rs group (ses 1 hrough 6) included 4 gohic novels, 4 hisorical novels, 4 naional ales,

    4 indusrial novels, 4 silver-ork novels, and 4 Bildungsromane. O he 6 ses in he second

    group, 3 were also presen in he rs (ses 8, 9, and 12: 2 exs each rom indusrial novels,

    gohic novels, and Bildungsromane), whereas he oher 3 were no (ses 7, 10, and 11: 2

    exs each rom ani-Jacobin, evangelical, and Newgae novels). Docuscopes ask was o

    nd and mach he 3 ses rom he second group ha were also presen in he rs.6

    5 We limied ourselves o his daabase because mos oher exs available on he web in 2006-8 appeared oo

    unreliable or our purposes. Today, our assessmen would be dieren, and a new iniial pool would probably modiy

    imporan aspecs o our research.

    6 This is he complee lis o he exs: se 1 (gohic novels): A Sicilian Romance, The Old Manor House, The Monk,and Melmoth the Wanderer; se 2 (hisorical novels): Waverley, Ivanhoe, The Entail, and Valperga; se 3 (naional

    ales): Castle Rackrent, The Wild Irish Girl, The Absentee, and Marriage; se 4 (indusrial novels): Shirley, Alton

    Locke, Hard Times, and North and South; se 5 (silver-ork novels): Glenarvon, Vivian Grey, Pelham, and Mrs Ar-

  • 7/30/2019 Literary Lab Pamphlet 1

    6/29

    4

    To be sure he wouldn unconsciously il his work on Docuscopes resuls in a pre-deer-

    mined direcion, Wimore asked o be old nohing abou he exs he was receiving; ile-

    pages were removed rom he les (hey oen provide giveaway clues ha are less iner-

    esing han he microlinguisic moves ha ge made in he ex), and he lierally walked

    ino he meeing wihou knowing how Docuscope had perormed. He was hoping ha

    Docuscope would ail a his es, he emailed us a ew days beore he meeing, since Ihave a sake in arguing ha i is maerial consrains on perormance (in plays) ha allows

    Docuscope o make inelligible genre discriminaions when i comes o Shakespeare. I

    Docuscope urns ou o be good a picking genres o novels as well, I am going o have o

    expand my noion o maerial consrain in is relaionship o language pracices. (Laer,

    hough, he seemed pleased a how well Docuscope had done.)

    Wimore used a variey o measures o mach he genres rom he wo groups. For ex-

    ample, he assessed he degree o which mulivariae saisical analysis could produce

    acors ha would pry apar pairs rom one anoher a acor being a patern o having

    cerain LATs and lacking cerain ohers.7

    He also compared each pairing agains a col-lecion o exs called he Frown Corpus (early 1990s American English) o see when hey

    boh exhibied idenical elevaed and depressed scores on LATs in comparison wih he

    average score rom Frown.8 By combining hese echniques, Wimore came up wih he

    ollowing maches: 2:9 (wih 1:9 a close second), 4:8, and 6:12. When he curain was lied,

    i urned ou ha Docuscopes only misake consised in mis-maching group 9 (gohic

    novels) wih group 2 (hisorical) raher han 1 (gohic): a mix-up mos lierary hisorians

    would consider venial, or maybe even ineviable, given he porous borders beween hese

    wo genres. (And hen, as Wimore wroe in his presenaion, he correc 1-9 pairing was

    indeed a close second.)

    mytage, or Female Domination; se 6 (Bildungsromane): Jane Eyre, The History o Pendennis, David Coppereld,

    and Daniel Deronda; se 7 (ani-Jacobin novels): Mordaunt, and Adeline Mowbray; se 8 (indusrial novels): The Lie

    and Adventures o Michael Armstrong,the Factory Boy, and Mary Barton; se 9 (gohic novels): The Mysteries o

    Udolpho, and Zofoya, or, The Moor; se 10 (evangelical novels): Coelebs in Search o a Wie, and Sel-Control; se

    11 (Newgae novels): Eugene Aramand Jack Sheppard; se 12 (Bildungsromane): Great Expectationsand Middle-

    march.

    Rerospecively, his lis is odd and awed in wo opposie ways. Firs, he 36 exs were chosen so as o maxi-

    mize variaion wihin each given genre. Alhough quie wrong as a way o selec a sample rom a populaion, his

    choice was mean o increase he severiy o he es: Docuscope had o prove i could recognize a genre even

    when given a quie disparae bundle o specimens. I his increased he difculy o he enerprise, a second deci-

    sion did he exac opposie: insead o giving Wimore 36 exs o be assigned o various generic classes, Moreti

    gave him discree groups ha were already subdivided ino genres. This, clearly, made maters much easier, as he

    inernal variaion wihin any given genre could be averaged ou by looking a he group as a whole.

    These odd, aniheical decisions show how unprepared we were as a group or should we say: as a discipline?

    or his ype o research. The idea o a random sample, or insance, never really crossed our minds...

    7 One can hink o a acor as a recipe or describing recurring paterns o variaion in a larger collecion o iems. I

    each novel is a sack o cards, Docuscope examines all o he decks and couns wha is in hem. Then acor analy-

    sis goes hrough all o he conens o each sack and says, whenever I see los o red sixes, I see very ew ours and

    ves o any kind. These recipes o presences and absences can hen be esed agains imposed groups o hose

    sacks (genres) o see i he acors reliably disinguish iems rom each.

    8 Use o a reerence corpus seemed like a good idea, and since Frown had been used o es Docuscope in is

    developmen, hose comparisons were buil ino he ool and so available or ready use. I urned ou ha he Frown

    comparisons were he mos accurae in predicing lierary criical genre judgmens.

  • 7/30/2019 Literary Lab Pamphlet 1

    7/29

    5

    As he meeing was nearing is end, John Bender asked he hard quesion ha was hang-

    ing in he air: Sriking as hese resuls were, did we hink hey had produced new knowl-

    edge? The answer, o course, was No: Docuscope had corroboraed wha lierary scholars

    already knew or a leas were convinced o i.e. ha cerain exs belonged o he same

    class. No new knowledge here. Bu ha human judgmen and unsupervised saisical

    analysis would agree on genre classicaion his wasa novely ha had emerged romhe es. Jus as Docuscope had corroboraed exising scholarship, he later had proven

    Docuscopes reliabiliy. We waned o know wheher i could replicae is Shakespeare

    resuls in unamiliar erriory, and i could; ha rs experimen had no been a uke. A

    compuer could classiy lierary exs. And when Wimore in passing, and almos as an

    aerhough showed an old, unpublished char rom his Shakespeare sudy, he pos-

    sibiliy seemed even richer in implicaions.

    3. March 2009: Mos Frequen Words Recognize Novelisic Genres

    Docuscope had passed he es. Was i he only program ha could do so? Mathew Jock-

    ers, who had been working on auhorship sudies or a while, waned o see wheher he

    mehods he had been developing could be applied o genre recogniion as well. In many

    ways, genre classicaion is akin o auhorship atribuion. Bu here is one imporan di-

    erence. Wih auhorship problems, one atemps o exrac a eaure se ha excludes

    conex-sensiive eaures rom he analysis, he consensus being ha a se made up

    primarily o requen, or closed-class, word eaures yields he mos accurae resuls. For

    genre classicaion, however, one would inuiively assume ha conex words say:

    casle in gohic novels would be criical. Ye, Jockerss preliminary resuls suggesed

    ha an equally disinc genre signal may be deeced rom a small se o high-requency

    eaures.

    Using jus 44 word and puncuaion eaures which we evenually ended up calling Mos

    Frequen Words, or MFW Jockers was able o classiy he novels in he corpus as well

    as Wimore had done wih Docuscope (and is ar more complex eaure se). 9 Using he

    dis and hclus uncions in he open-source R10 saisics applicaion, Jockers clus-

    ered he exs in he dendrogram o gure 3.1:

    9 To derive his eaure se, Jockers lowercased he exs, couned and convered o relaive requencies he vari-

    ous eaure ypes, and hen winnowed he eaure se by choosing only hose eaures ha have a mean relaive

    requency o .03% or greaer. This resuled in a marix consising o he ollowing 44 eaures (he prex p_ in-

    dicaes a puncuaion oken ype insead o a word oken): a, all, and, as, a, be, bu, by, or, rom,

    had, have, he, her, him, his, i, in, is, i, me, my, no, o, on, p_apos, p_comma, p_exlam,

    p_hyphen, p_period, p_ques, p_quoe, p_semi, said, she, so, ha, he, his, o, was, which,wih, you.

    10 htp://www.r-projec.org/

  • 7/30/2019 Literary Lab Pamphlet 1

    8/29

    6

    National

    Industrial1

    Industrial2

    Newgate

    Silver-Fork

    Bildungsroman1

    Bildungsroman2

    Evangelical

    AntiJacobin

    Gothic2

    Gothic1

    Historical

    Plot Created: Feb. 9, 2009

    By: mjockers

    Novelistic Genres

    Using Euclidean Distance with Complete Linkage and 42 Features

    Figure 3.1: Cluser Dendrogram o novel genres using Mos Frequen Words (MFW).

    Aer Jockers shared his resuls wih Wimore, Wimore suggesed esing his mehodol-

    ogy on he Shakespeare corpus. Once again, MFW accuraely clusered he majoriy o

    Shakespeares plays ino he ragedies, comedies, hisories, and lae plays o gure

    3.2.

    Quaniaive Formalism, reads he ile o his aricle. Formalism, because all o us, in one

    way or anoher, were ineresed in he ormal convenions o genre; and quaniaive, be-

    cause we were looking or more precise ideally, measurable ways o esablish generic

    dierences. So, we really waned Docuscope and MFW o do well. Bu so well, no one had

    hough possible: no only were genre signals quie srong hey were equally srong a

    wholly diferen exual levels: jus as recognizable by Docuscopes mix o grammar and

    semanics, as by he handul o uncion words o MFW. The convergence was so clear, i

    was almos spooky: i suggesed ha he logic o genre reached a deph ha no one had

    imagined, and no one really knew how o explain. The requency o aricles and conjunc-

    ions which allowed he idenicaion o Newgae novels or Bildungsromanein ex aer

    ex could his really be essenial o he uncioning o a genre? Why?

  • 7/30/2019 Literary Lab Pamphlet 1

    9/29

    7

    Comedy_AllsWell

    Comedy_Merchant

    Comedy_Measure

    Comedy_AsYouComedy_MuchAdo

    Comedy_Errors

    Comedy_Two Gentlemen

    Comedy_TwelhNight

    Tragedy_Othello

    Comedy_MerryWives

    Comedy_Taming

    History_JOHN

    History_1HENRYVI

    History_RICHARDII

    History_HENRYV

    History_2HENRYIV

    History_HENRYIV

    Late_HenryVIII

    Tragedy_Hamlet

    Tragedy_Titus

    History_RICHARDIII

    History_2HENRYVIIHistory_3HENRYVI

    Tragedy_Macbeth

    Tragedy_Coriolanus

    Late_Cymbeline

    Late_Winters

    Comedy_LoveLabours

    Comedy_Midsummer

    Tragedy_Julius

    Tragedy_Romeo

    Tragedy_Troilus

    Late_Tempest

    Tragedy_Timon

    Tragedy_Antony

    Tragedy_Lear

    Plot Created: Feb. 4, 2009

    By: mjockers

    Shakespeare Plays

    Using Euclidean Distance with Complete Linkage and 37 Features

    Figure 3.2: Dendrogram o Shakespeare Firs Folio plays using Mos Frequen Words wih major clusers highligh-

    ed. Here Jockers used he 37 eaures rom he Shakespeare plays ha had a mean relaive requency o greaer han

    or equal o .03%. Noe he similariy beween his ree and Docuscopes diagram in g. 1.1, wih he close pairings o

    Winters Taleand Cymbeline; 2 Henry VIand 3 Henry VI, and he proximiy oCoriolanuso he Cymbeline-Winters

    Talepair.

    As soon as school was over, we me again.

  • 7/30/2019 Literary Lab Pamphlet 1

    10/29

    8

    4. June 2009: Forking Pahs

    Our nex meeing, a Sanord, began wih Wimore showing a page ha Docuscope had

    isolaed as he mos gohic o he enire corpus ha is o say, he one which presened

    an exremely high number o ypically gohic eaures (gure 4.1):

    Figure 4.1: Docuscope screensho o okens diereniaing he gohic rom several oher genres, drawn rom Ann

    Radclie, A Sicilian Romance(1790). These diereniaing bundles o LATs were idenied hrough acor analysis

    and ANOVA, wih acors winnowed hrough he Tukey es.

    I was an ineresing momen; no jus because he idea o a genres ypical page was un-

    usual and inriguing, bu because, as Sarah Allison immediaely poined ou, he gohic o

    Docuscope appeared o be quie dieren rom ha o Humanscope (as she called i): i

    was no he same gohic wesaw. For us, ha page was gohic because o he subdued er-

    ror and he archway, he ruin and apprehension and he limbs ha rembled no because

    o he he him his had was sruck he and heard he which caugh Docuscopesatenion. Beween he wo approaches, here seemed o be nohing in common. Or per-

    haps, more precisely: nohing in common, in erms o heir unis o analysis;bu everyhing

    in common in erms o resuls: wheher via banditi and blood, or utered he and covered

    him, Humanscope and Docuscope agreed ha his page belonged o he gohic, and o

    no oher genre. And a his poin, he idea ha had rs conusedly crossed our minds a

    ew monhs earlier crysallized once and or all: genres, like buildings, possess disincive

    eaures a every possible scale o analysis: morar, bricks, and archiecure, as Ryan Heus-

    er, pu i: he morar, he grains o sand, o Mos Frequen Words, he bricks o Docuscopes

    lexico-grammaical caegories, and he archiecure o hemes and episodes ha readers

    recognize. The hree layers were no even overlapping; heir signals were largely disincrom each oher. Dieren as he hree layers were among hemselves, hough, hey were

    also dieren rom he corresponding layers o oher genres: he gohic morar oally

  • 7/30/2019 Literary Lab Pamphlet 1

    11/29

    9

    unlike he morar o he naional ale, or he ani-Jacobin novel; he gohic bricks unlike

    he bricks used by oher genres, and he same or he more visible archiecural shapes.

    We will reurn o he concepual quesions posed by hese observaions owards he end

    o his aricle. On ha day in June, hough, somehing else seemed even more inspiring:

    he char we briey menioned a he end o secion 2, which displayed all o Shakespeare

    plays along wo orhogonal axes (gure 4.2: Shakespeares Plays)

    AD

    DA

    A

    A

    A

    AA

    B DD

    DA

    AD

    BB

    B

    DDB

    B D DBCC A

    A

    D

    BA

    C

    B

    Shakespeares Plays

    PC2

    PC1

    1.5

    1

    0.5

    0

    -0.5

    -1

    -1.5

    -2

    -2 -1 0 1 2

    Figure 4.2: Scaterplo marix in which Shakespeares plays are raed on heir rs wo principal componens aer

    having been couned by Docuscope and analyzed in erms o aggregaes o LATs. PCA perormed on he covari-ance marix, unscaled daa. Iem key: A = comedy, B = Hisory, C = Lae Plays, D = Tragedies. Noe how he wo

    componens place comedies in he upper righ quadran, hisories in he lower le, and several lae plays in he

    lower righ (whereas ragedies, or some reason, are dispersed all over he eld).

    Wimore and Hope had abandoned he idea o publishing his diagram in a scholarly book

    o radiional lierary criicism: hey el i would be more eecive o make heir poin en-

    irely wih words. Bu he group saw in he char he promise o an inuiive, synheic view

    o he lierary eld, wih each genre placed in relaion o all he ohers. Moreti, in paricular,

    was sruck by he similariy beween he char and he principal componens chars ha

    Cavalli-Sorza (e.al.), in The Hisory and Geography o Human Genes, had used o racerelaionships among human populaions.11 Could narraive genres be similarly reduced o

    wo basic variables? And would he ensuing disribuion correlae wih, say, Bourdieus

    11 See L. Luca Cavalli-Sorza, Paolo Menozzi, and Albero Piazza, The History and Geography o Human Genes,

    Princeon UP 1994, especially pp. 39. Principal componen analysis is a procedure, similar o acor analysis, which

    reduces he variance exising wihin a group o objecs -- in our case, he linguisic-sylisic dierence among lier-

    ary exs -- o wo orhogonal axes, called Principal Componen 1 and 2 (PC1 and PC2). Principal Componen 1 is he

    combinaion o eaures ha expresses he maximum amoun o variance available o a single componen; Principal

    Componen 2 displays a urher increase o variance orhogonally wih respec o PC1. Taken ogeher, PC1 and PC2

    are a very economical way o represening as much variance as i is possible on wo dimensions; however, heynever express he oal amoun o variance wihin a sysem, bu, raher, a rade-o beween high inuiive visibiliy

    and a (limied) loss o precision.

    Shakespeares Plays

  • 7/30/2019 Literary Lab Pamphlet 1

    12/29

    10

    sociological (bu highly subjecive) map o he French lierary eld? Could we acually

    map morphology over social disincion?

    Wimores char seemed perec or all his. Even he ac ha i wasn perec wih hose

    ragedies udging he more orderly paterns o he oher genres seemed a sign o reli-

    abiliy, as hisory is isel never perec. So, we decided o repea he atemp wih novelis-

    ic genres. I he resuls were good, wo urher developmens would become imaginable.

    Firs, he sysem o genres migh urn rom a hodge-podge o unrelaed caegories12 o

    a single marix o inerconneced ormal variables. And, second, i migh become pos-

    sible o char he Grea Unread he vas, unexplored archive ha lies underneah he

    narrow canon o lierary hisory. One could give Docuscope and MFW housands o exs

    o unknown generic afliaion, and see where hey would all in he graviaional eld o

    beter-known genres. One could envisage generaion-by-generaion maps o he lierary

    universe, wih galaxies, supernovae, black holes ...

    Wih hese quesions running hrough our heads, we re-deployed he February and March

    daa along he lines o gure 4.2. The rs visualizaion, produced by MFW gure 4.3 urned ou o be perecly ambiguous: promising and perplexing in equal measure. There

    was cerainly less clariy han in he Shakespeare case; bu, we were charing wice as many

    genres, and over a much longer period. And hen, some paterns werevisible: wih a ew

    excepions, gohic and hisorical novels lay on he negaive side o principal componen

    1 (he le side o he horizonal axis), while he Bildungsromanand indusrial novels were

    clearly on is posiive side. For us, his was boh good and bad news. Good, because a

    patern is wha one always looks or, in exploraory work. Bu bad, because he patern was

    chronological, more han ormal: one generaion, hen a second, more conused one, and

    hen a hird. Was principal componen 1 capuring genresignals hen or hisorical ones?

    The later seemed more likely, especially given how poorly hose genres ha ourished

    in he same years (gohic/hisorical; silver-ork/Newgae; indusrial/Bildungsroman) were

    separaed. Hisory seemed deniely sronger han orm.

    Bu here were also some daa ha conradiced he hisorical alignmen: in he crowded

    cenral secion, which conained genres rom wo dieren generaions, he verical axis

    o PC2 which separaed ani-Jacobin and evangelical novels rom Newgae sories

    migh be capuring genre signals aer all.13 Would i be possible o isolae such signals,

    and magniy hem?

    12 Righ now, he very names o novelisic genres are a elling even maddening sign o caegorical conusion

    highlighing now he novels medium (he episolary novel), now is conen (hisorical, indusrial), syle (nauralis),

    proagonis (picaresque, pasoral), all he way o more or less anciul meaphors (gohic, silver-ork).

    13 Then again, wih only wo exs each or hese genres, his could easily be he resul o chance. Or no.

  • 7/30/2019 Literary Lab Pamphlet 1

    13/29

    11

    A

    A

    B

    B

    B

    B

    B

    E

    E

    G

    G G

    G

    G

    G

    H

    H

    H

    H

    I

    I

    I

    I

    I

    N

    N

    N

    N

    K

    K

    S

    S

    S

    S

    3 2 1

    2

    1

    PC1

    PC2

    All Novels Using 51 Features

    0

    1

    2

    0 1 2

    Figure 4.3: A graphical represenaion o he rs wo principal componens in a PCA analysis o he Mos FrequenWords (MFW). Each leter represens a single ex (A=ani-Jacobin novels, B=Bildungsromane, E=evangelical nov-

    els, G=gohic novels, I=indusrial novels, K=Newgae novels, N=naional ales, S=silver-ork novels).

    5. June-Sepember 2009: Dead End

    From June o Sepember, Wimore and Jockers kep looking or ways o improve he early

    resuls o PC analysis. Firs, hey segmened he exs o see wheher smaller unis would

    improve diereniaion. All exs were divided ino en equal pars bu he resuls did nochange much. Then, noicing ha he segmens disribuion was oen very uneven as

    in gure 5.1, where abou one hird o hem udge an oherwise good separaion beween

    gohic and hisorical novels we decided o label all he segmens: Hisorical.8.1 would

    indicae he rs segmen o Windsor Casle(which happened o be he eighh hisorical

    novel in our corpus); Gohic.1.10 he enh segmen o Vahek(which was he rs gohic

    ex), and so on. The overlap among dieren genres migh urn ou o be limied o spe-

    cic porions o he exs (beginnings, or endings); i ha were so, and genres became

    more disincive more hemselves, as i were a specic momens in he plo, hen one

    could ocus on hose momens and magniy heir separaion. I was a plausible, perhaps

    even an ingenious hypohesis. Bu no. Some novels were mos disincive early on; oh-ers, lae in he plo; or in he middle; or nowhere in paricular.

  • 7/30/2019 Literary Lab Pamphlet 1

    14/29

    12

    +

    +

    +

    +

    ++

    +

    +

    +

    ++

    ++

    +

    +

    +

    +

    +

    ++

    +

    +

    +

    +

    ++

    +

    +++

    +++

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    ++

    +

    +

    ++ + +

    +

    +

    +

    ++

    +

    +

    +

    +

    ++

    +

    +

    ++

    +

    +

    +

    +

    +

    +

    +

    +

    +

    Group

    Gothic

    Historical

    -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

    6

    5

    4

    3

    2

    1

    0

    -1

    -2

    -3

    -4

    -5

    -6

    PC2

    PC1

    36 Novels

    Figure 5.1: 8000-word segmens o he rs wo groups o 36 novels, raed by Docuscope on rs wo principal

    componens. In all PCA analyses below, daa are scaled (i.e., PCA is perormed on he correlaion marix o percen-

    age scores).

    Nex, we urned o he composiion o our corpus: as explained in oonoe 6, he iniial

    collecion o 36 exs ended o exaggerae variaion wihin each genre, making lie unnec-

    essarily hard or Docuscope and MFW. We reurned o he Chadwyck-Healey daabase

    and added o he iniial corpus all hose exs ha exising bibliographies had assigned o

    specic genres; included wo new genres (Jacobin and sensaion novels); and repeaed

    all he calculaions on he new corpus o 106 exs.14

    Nohing.

    14 This second corpus also included a ew exs, mosly rom minor genres, scanned or us by he Sanord librar-ies. Since however he Chadwyck-Healey daabase remained he major source, canonical exs sill predominaed:

    o 28 hisorical novels, or insance, 14 were by Scot.

    36 Novels

  • 7/30/2019 Literary Lab Pamphlet 1

    15/29

    13

    Maybe rying o char eigh decades a once was oo much. We divided he corpus ino

    hree generaions;15 hough o course less crowded, he new chars were jus as indeci-

    sive. By he end o summer, i was clear ha he resuls were no longer changing.

    6. November 2009: Auhors vs. Genres

    In November, in he course o a eleconerence which included he ve auhors and a ew

    Sanord grad sudens, we looked again a he 3 generaional maps, which now included

    all individual exs (gures 6.1-3), and all o a sudden realized how srong he auhor sig-

    nal was. Remember, we didn wan auhors; we waned genres. Bu i was impossible no

    o noice ha Docuscope and MFW clusered he ormer much beter han he later. Wih

    Dickens, Bron, and Elio, or insance who had all writen boh indusrial novels and Bil-

    dungsroman he pull o he auhor in gure 6.3 was clearly much sronger han ha o

    he genre; and he same was rue or Bulwer-Lytons Las Days o Pompeii, Eugene Aram,and Pelham, closely clusered ogeher in gure 6.2, despie he ac ha hey belonged o

    he raher dieren genres o hisorical, Newgae, and silver-ork cion.

    PC1

    PC2

    4

    2

    0

    2

    4 2

    Genres

    Five Genres in PCA Space

    Anti-Jacobin

    Evangelical

    GothicJacobinNational

    0 2 4

    ShelleySmith

    RadcliffeInchbald

    Morgan

    Edgeworth

    Figures 6.1-3: Generaional analysis o original 36 novels as raed by Docuscope on rs wo principal componens.

    Noice he proximiy among he exs by Inchbald, Smih, Radclie, Shelley, Morgan, and Edgeworh in 6.1; by Ain-

    sworh, Porer, Lyton, Gal, and o course Scot, in 6.2; by Gaskell, Dickens, Bron, Collins and Elio in 6.3.

    15 The rs generaion (ca. 1790-1820) included gohic, Jacobin, ani-Jacobin, naional ales, and evangelical nov-els; he second (ca. 1815-1850) hisorical, silver-ork, and Newgae novels; he hird (ca. 1845-1875) indusrial, Bil-

    dungsroman, and sensaion novels.

  • 7/30/2019 Literary Lab Pamphlet 1

    16/29

    14

    GenresHistorical

    Newgate

    Silver Fork

    Three Genres in PCA Space

    PC1

    PC2

    4

    2

    0

    2

    4

    4 2 0 2 4

    Porter

    Lyon

    Ainsworth

    Sco

    Figure 6.2: (see capion above)

    GenresBildungsroman

    Industrial

    Sensation

    Three Genres in PCA Space

    PC1

    PC2

    3

    2

    1

    0

    1

    2

    3

    2

    Collins

    Eliot

    Bronte

    Dickens

    Gaskell

    0 2 4

    Figures 6.2-3: (See capion or 6.1)

    Why should auhors be so much more recognizable han genres? Probably, because

    Docuscope and MFW are very good a capuring somehing all wriers do, wheher heyknow i or no: using impercepible linguisic paterns ha provide an unmisakable sy-

    lisic signaure. Genres also have such sylisic signaures, o course; bu genres have a

  • 7/30/2019 Literary Lab Pamphlet 1

    17/29

    15

    narraivesignaure oo heir plo which is a leas as imporan. The episodes ha so

    powerully ideniy he Bildungsromanor insance discussions wih old menors and

    young riends, alse sars, disappoinmens, he discovery o ones vocaion ... all his

    has no equivalen in a sensaion novel; jus as a sensaion novels myseries and murders

    would make no sense in an indusrial novel, and so on. So, wha happens when he same

    wrier moves rom one genre o anoher when, say, Dickens moves rom he indusrialnovel Hard Timeso he urban muliplo o Litle Dorri, he hisorical Tale o Two Ciies, or

    he Bildungsromano Grea Expecaions wha happens is ha his plos change, buhis

    syle doesn. Or no as much. The sories o Cokeown, London, or Paris are much more

    dieren han he words Dickens uses o narrae hem. His language remains basically he

    same.

    Why did Docuscope and MFW recognize auhors so well, hen and genres less well? Be-

    cause hey had been designed o recognize language, bu no plo.16 They were probably

    doing he bes ha could be done in separaing genres on he sole basis o heir language

    and syle; bu language and syle are jus no enough o delimi a genre rom anoher. Andaer all, why should hey be? In addressing heir readers, genres use boh syle and plo

    (in he nineeenh cenury, probably, more plo han syle): our programs were missing

    hal o he srucure, and i made sense ha hey should be only hal successul. Hal suc-

    cessul does no mean un-successul. Bu i does sugges ha an analyical ool capable

    o quaniy plo is sill missing.17 And as long as ha is he case, he generic disribuion

    eeced by Docuscope and MFW was oo random o suppor a good lierary axonomy,

    le alone an exploraion o he archive. The Grea Unread would, or he ime being, remain

    unread.

    7. December 2009: 220 Chars

    In December, Allison, Heuser, and Moreti urned o a new se o visualizaions: wo series

    o chars ha included all possible pairings among he 11 genres o he enlarged cor-

    pus (gohic/Jacobin, gohic/ani-Jacobin, gohic/naional ale, and so on, all he way o

    he oher end o he chronological specrum). These chars came in wo orms; he rs

    showed he disribuion o wo genres based on MFW (gure 7.1) and Docuscope (g-

    ure 7.2). These were our basic ools, allowing us o inuiively grasp wheher wo specic

    genres separaed well as gohic and sensaion novels in gures 7.1-2 or no. (MFW andDocuscope, incidenally, urned ou o be equally able or unable, as he case may be o

    separae genres rom each oher.)

    16 They can cerainly see how acions are described: wih simple or complex senences, sressing subjecive mood

    or objecive resuls, surprise or rerospecion. Bu hey can hardly see wha acions consis o: a sorys chrono-

    logical (and semanic) chain largely eludes hem.

    17 This nding cheered Wimore, since i suggess ha in novelisic represenaion, plo provides an avenue o

    generic diereniaion ha has to beless visible o Docuscope because i does no have o be ied o he physical

    limis o he medium, whereas Renaissance drama consanly grappling wih he difculy o elling sories wihreal bodies in a ew hours migh have his exra-sylisic avenue oreclosed, leading o more legible (because

    maerially consrained) generic syles a he level o he senence.

  • 7/30/2019 Literary Lab Pamphlet 1

    18/29

    16

    PC1

    PC2

    4

    2

    0

    2

    4

    Genres

    Gothic

    Sensation

    5 0 5

    Two Genres in PCA Space

    Figure 7.1: Mos Frequen Word scater plo. Here, and in all oher PCA chars, each poin (circle or riangle) on he

    plo sands or one segmen (one enh) o a ex.

    PC1

    PC2 Genres

    Gothic

    Sensation

    Two Genres in PCA Space

    2

    0

    2

    4

    6

    6 4 2 0 2 4

    Figure 7.2: Docuscope scater plo.

    The second ype o char re-deployed he circles and riangles o gures 7.1-2 add-

    ing wo urher eaures. Firs, i agged each segmen, making explici which (par o

    which) ex i came rom: he circles in he lower righ corner o gure 7.1, or insance,

    urned ou in gure 7.3 o belong o Vahek, hus bringing o ligh he cenraliy

    or eccenriciy, as he case may be o each ex wihin is genre (an issue which

  • 7/30/2019 Literary Lab Pamphlet 1

    19/29

    17

    may have proound consequences or our knowledge o genre, and which we plan

    o invesigae in he uure). And hen, gures 7.3-4 also indicaed which rais o he

    wo principal componens conribued o he specic shape o a genres disribu-

    ion: which words, or Docuscope Dimensions exered a sronger pull in separaing

    gohic rom sensaion novels. So, or insance, he lower righ quadran o gure 7.3

    highlighs he denie aricle as an imporan dierenial eaure o he gohic in MFWanalysis (compare wih gure 7.1); in gure 7.4, a similar role is played, in he lower le

    quadran, by Narraive VP, Pronouns, and Reporing Evens (compare wih gure

    7.2).18

    5

    5

    0

    5

    PC1

    PC2

    Goth_01_1_1786_Beckf_VathekTran

    Goth_01_10_1786_Beckf_VathekTran

    Goth_01_2_1786_Beckf_VathekTran

    Goth_01_3_1786_Beckf_VathekTran

    Goth_01_4_1786_Beckf_VathekTran

    Goth_01_5_1786_Beckf_VathekTranGoth_01_6_1786_Beckf_VathekTran

    Goth_01_7_1786_Beckf_VathekTran

    Goth_01_8_1786_Beckf_VathekTran

    Goth_01_9_1786_Beckf_VathekTran

    Goth_02_1_1788_Smith_EmmelinethGoth_02_10_1788_Smith_Emmelineth

    Goth_02_2_1788_Smith_Emmelineth

    Goth_02_3_1788_Smith_Emmelineth

    Goth_02_4_1788_Smith_Emmelineth

    Goth_02_5_1788_Smith_Emmelineth

    Goth_02_6_1788_Smith_Emmelineth

    Goth_02_7_1788_Smith_Emmelineth

    Goth_02_8_1788_Smith_Emmelineth

    Goth_02_9_1788_Smith_Emmelineth

    Goth_03_1_1790_Radcl_AS

    Goth_03_10_1790_Radcl_ASicilianR

    Goth_03_2_1790_Radcl_ASicilianRGoth_03_3_1790_Radcl_ASicilianR

    Goth_03_4_1790_Radcl_ASicilianR

    Goth_03_5_1790_Radcl_A

    Goth_03_6_1790_Radcl_ASicilianR

    Goth_03_7_1790_Radcl_ASicilianR

    Goth_03_8_1790_Radcl_ASicilianRGoth_03_9_1790_Radcl_ASicilianR

    Goth_04_1_1791_Radcl_TheRomance

    Goth_04_10_1791_Radcl_TheRomance

    Goth_04_2_1791_Radcl_TheRomance

    Goth_04_3_1791_Radcl_TheRomance

    Goth_04_4_1791_Radcl_TheRomance

    Goth_04_5_1791_Radcl_TheRomance

    Goth_04_6_1791_Radcl_TheRomance

    Goth_04_7_1791_Radcl_TheRomance

    Goth_04_8_1791_Radcl_TheRomance

    Goth_04_9_1791_Radcl_TheRomance

    Goth_05_1_1794_Radcl_TheMysteri

    Goth_05_10_1794_Radcl_TheMysteriGoth_05_2_1794_Radcl_TheMysteri

    Goth_05_3_1794_Radcl_TheMysteri

    Goth_05_4_1794_Radcl_TheMysteri

    Goth_05_5_1794_Radcl_TheMysteri

    Goth_05_6_1794_Radcl_TheMysteri

    Goth_05_7_1794_Radcl_TheMysteri

    Goth_05_8_1794_Radcl_TheMysteri

    Goth_05_9_1794_Radcl_TheMysteriGoth_06_1_1796_Lewis_TheMonkARo

    Goth_06_10_1796_Lewis_TheMonkARo

    Goth_06_2_1796_Lewis_TheMonkARo

    Goth_06_3_1796_Lewis_TheMonkARo

    Goth_06_4_1796_Lewis_TheMonkARo

    Goth_06_5_1796_Lewis_TheMonkARo

    Goth_06_6_1796_Lewis_TheMonkARo

    Goth_06_7_1796_Lewis_TheMonkARo

    Goth_06_8_1796_Lewis_TheMonkARo

    Goth_06_9_1796_Lewis_TheMonkARo

    Goth_07_1_1797_Radcl_TheItalian

    Goth_07_10_1797_Radcl_TheItalian

    Goth_07_2_1797_Radcl_TheItalian

    Goth_07_3_1797_Radcl_TheItalian

    Goth_07_4_1797_Radcl_TheItalian

    Goth_07_5_1797_Radcl_TheItalianGoth_07_6_1797_Radcl_TheItalian

    Goth_07_7_1797_Radcl_TheItalian

    Goth_07_8_1797_Radcl_TheItalian

    Goth_07_9_1797_Radcl_TheItalian

    Goth_08_1_1799_Godwi_StLeonATal

    Goth_08_10_1799_Godwi_StLeonATalGoth_08_2_1799_Godwi_StLeonATal

    Goth_08_3_1799_Godwi_StLeonATal

    Goth_08_4_1799_Godwi_StLeonATal

    Goth_08_5_1799_Godwi_StLeonATal

    Goth_08_6_1799_Godwi_StLeonATal

    Goth_08_7_1799_Godwi_StLeonATal

    Goth_08_8_1799_Godwi_StLeonATal

    Goth_08_9_1799_Godwi_StLeonATal

    Goth_09_1_1806_Dacre_ZofloyaorT

    Goth_09_10_1806_Dacre_ZofloyaorT

    Goth_09_2_1806_Dacre_ZofloyaorT

    Goth_09_3_1806_Dacre_ZofloyaorT

    Goth_09_4_1806_Dacre_ZofloyaorT

    Goth_09_5_1806_Dacre_ZofloyaorT

    Goth_09_6_1806_Dacre_ZofloyaorT

    Goth_09_7_1806_Dacre_ZofloyaorT

    Goth_09_8_1806_Dacre_ZofloyaorT

    Goth_09_9_1806_Dacre_ZofloyaorT

    Goth_10_1_1810_Shell_ZastrozziAGoth_10_10_1810_Shell_ZastrozziA

    Goth_10_2_1810_Shell_ZastrozziA

    Goth_10_3_1810_Shell_ZastrozziA Goth_10_4_1810_Shell_ZastrozziA

    Goth_10_5_1810_Shell_ZastrozziA

    Goth_10_6_1810_Shell_ZastrozziA

    Goth_10_7_1810_Shell_ZastrozziAGoth_10_8_1810_Shell_ZastrozziA

    Goth_10_9_1810_Shell_ZastrozziA

    Goth_11_1_1820_Matur_Melmoththe

    Goth_11_10_1820_Matur_MelmoththeGoth_11_2_1820_Matur_Melmoththe

    Goth_11_3_1820_Matur_Melmoththe

    Goth_11_4_1820_Matur_Melmoththe

    Goth_11_5_1820_Matur_Melmoththe

    Goth_11_6_1820_Matur_Melmoththe

    Goth_11_7_1820_Matur_Melmoththe

    Goth_11_8_1820_Matur_Melmoththe

    Goth_11_9_1820_Matur_Melmoththe

    Goth_12_1_1818_Shell_Frankenste

    Goth_12_10_1818_Shell_Frankenste

    Goth_12_2_1818_Shell_Frankenste

    Goth_12_3_1818_Shell_FrankensteGoth_12_4_1818_Shell_Frankenste

    Goth_12_5_1818_Shell_Frankenste

    Goth_12_6_1818_Shell_Frankenste

    Goth_12_7_1818_Shell_Frankenste

    Goth_12_8_1818_Shell_Frankenste

    Goth_12_9_1818_Shell_Frankenste

    Goth_13_1_1793_Smith_TheOldMano

    Goth_13_10_1793_Smith_TheOldMano

    Goth_13_2_1793_Smith_TheOldManoGoth_13_3_1793_Smith_TheOldMano

    Goth_13_4_1793_Smith_TheOldMano

    Goth_13_5_1793_Smith_TheOldMano

    Goth_13_6_1793_Smith_TheOldManoGoth_13_7_1793_Smith_TheOldMano

    Goth_13_8_1793_Smith_TheOldMano

    Goth_13_9_1793_Smith_TheOldMano

    Sens_01_1_1862_Bradd_LadyAudley

    Sens_01_10_1862_Bradd_LadyAudley

    Sens_01_2_1862_Bradd_LadyAudley

    Sens_01_3_1862_Bradd_LadyAudley

    Sens_01_4_1862_Bradd_LadyAudley

    Sens_01_5_1862_Bradd_LadyAudley

    Sens_01_6_1862_Bradd_LadyAudley

    Sens_01_7_1862_Bradd_LadyAudleySens_01_8_1862_Bradd_LadyAudley

    Sens_01_9_1862_Bradd_LadyAudley

    Sens_02_1_1868_Colli_TheMoonsto

    Sens_02_10_1868_Colli_TheMoonsto

    Sens_02_2_1868_Colli_TheMoonsto

    Sens_02_3_1868_Colli_TheMoonsto

    Sens_02_4_1868_Colli_TheMoonsto

    Sens_02_5_1868_Colli_TheMoonsto

    Sens_02_6_1868_Colli_TheMoonsto

    Sens_02_7_1868_Colli_TheMoonsto

    02_8_1868_Colli_TheMoonsto

    _02_9_1868_Colli_TheMoonsto

    Sens_03_1_1866_Colli_ArmadaleBy

    Sens_03_10_1866_Colli_ArmadaleBy

    Sens_03_2_1866_Colli_ArmadaleBy

    Sens_03_3_1866_Colli_ArmadaleBy

    Sens_03_4_1866_Colli_ArmadaleBy

    Sens_03_5_1866_Colli_ArmadaleBy

    Sens_03_6_1866_Colli_ArmadaleBy

    ns_03_7_1866_Colli_ArmadaleBy

    Sens_03_8_1866_Colli_ArmadaleBy

    Sens_03_9_1866_Colli_ArmadaleBy

    Sens_04_1_1859_Colli_TheWomanin

    Sens_04_10_1859_Colli_TheWomanin

    Sens_04_2_1859_Colli_TheWomanin

    Sens_04_3_1859_Colli_TheWomanin

    Sens_04_4_1859_Colli_TheWomanin

    Sens_04_5_1859_Colli_TheWomaninSens_04_6_1859_Colli_TheWomanin

    Sens_04_7_1859_Colli_TheWomanin

    Sens_04_8_1859_Colli_TheWomanin

    Sens_04_9_1859_Colli_TheWomanin

    Sens_05_1_1861_WoodH_EastLynneB

    Sens_05_10_1861_WoodH_EastLynneB

    Sens_05_2_1861_WoodH_EastLynneB

    Sens_05_3_1861_WoodH_EastLynneB

    Sens_05_4_1861_WoodH_EastLynneBSens_05_5_1861_WoodH_EastLynneB

    _1861_WoodH_EastLynneB

    Sens_05_7_1861_WoodH_EastLynneB

    Sens_05_8_1861_WoodH_EastLynneB

    Sens_05_9_1861_WoodH_EastLynneB

    0.3 0.2 0.1 0.0 0.1 0.2 0.3

    0.3

    0.2

    0.1

    0.0

    0.1

    0.2

    0.3

    a

    all

    an

    andas

    at

    be

    but

    by

    for

    from

    hadhave

    he

    her

    him

    his

    i

    if

    in

    is

    it

    memy

    no

    not

    of

    on

    or

    p_apos

    p_comma

    p_exlam

    p_hyphen

    p_period

    p_ques

    p_quote

    p_semi

    said

    she

    so

    that

    the

    they

    this

    to

    was

    were

    what

    when

    which

    who

    will

    with

    would

    you

    your

    0 5

    Figure 7.3: Mos Frequen Word scaterplo (ligh grey iles) and componen loadings (black).

    18 As each o he 55 genre pairings appeared in his double orm, we examined 110 chars produced by Docus-

    cope, and 110 produced by MFW. The mapping echnique used in gs. 7.3-4, in which dierenial rais becomevisible wihin he disribuion o he daa hemselves, is described in Mick Al, Exploring Hyperspace: A Non-Math-

    ematical Explanation o Multivariate Analysis, McGraw-Hill, London-NY 1990, chaper 4.

  • 7/30/2019 Literary Lab Pamphlet 1

    20/29

    18

    Goth_01_10_1786_Beckf_VathekTran.txt

    Goth_01_1_1786_Beckf_VathekTran.txt

    Goth_01_2_1786_Beckf_VathekTran.txt

    Goth_01_3_1786_Beckf_VathekTran.txt

    Goth_01_4_1786_Beckf_VathekTran.txt

    Goth_01_5_1786_Beckf_VathekTran.txtGoth_01_6_1786_Beckf_VathekTran.txt

    Goth_01_7_1786_Beckf_VathekTran.txt

    Goth_01_8_1786_Beckf_VathekTran.txt

    Goth_01_9_1786_Beckf_VathekTran.txt

    Goth_02_10_1788_Smith_Emmelineth.txtGoth_02_1_1788_Smith_Emmelineth.txt

    Goth_02_2_1788_Smith_Emmelineth.txt

    Goth_02_3_1788_Smith_Emmelineth.txtGoth_02_4_1788_Smith_Emmelineth.txt

    Goth_02_5_1788_Smith_Emmelineth.txt

    Goth_02_6_1788_Smith_Emmelineth.txt

    Goth_02_7_1788_Smith_Emmelineth.txt

    Goth_02_8_1788_Smith_Emmelineth.txt

    Goth_02_9_1788_Smith_Emmelineth.txt

    Goth_03_10_1790_Radcl_ASicilianR.txt

    Goth_03_1_1790_Radcl_ASicilianR.txt

    Goth_03_2_1790_Radcl_ASicilianR.txt

    Goth_03_3_1790_Radcl_ASicilianR.txt

    Goth_03_4_1790_Radcl_ASicilianR.txt

    Goth_03_5_1790_Radcl_ASicilianR.txt

    Goth_03_6_1790_Radcl_ASicilianR.txt

    Goth_03_7_1790_Radcl_ASicilianR.txt

    Goth_03_8_1790_Radcl_ASicilianR.txtGoth_03_9_1790_Radcl_ASicilianR.txt

    Goth_04_10_1791_Radcl_TheRomance.txt

    Goth_04_1_1791_Radcl_TheRomance.txt

    Goth_04_2_1791_Radcl_TheRomance.txt

    Goth_04_3_1791_Radcl_TheRomance.txtGoth_04_4_1791_Radcl_TheRomance.txt

    Goth_04_5_1791_Radcl_TheRomance.txt

    Goth_04_6_1791_Radcl_TheRomance.txt

    Goth_04_7_1791_Radcl_TheRomance.txt

    Goth_04_8_1791_Radcl_TheRomance.txt

    Goth_04_9_1791_Radcl_TheRomance.txtGoth_05_10_1794_Radcl_TheMysteri.txt

    Goth_05_1_1794_Radcl_TheMysteri.txt

    Goth_05_2_1794_Radcl_TheMysteri.txt

    Goth_05_3_1794_Radcl_TheMysteri.txt

    Goth_05_4_1794_Radcl_TheMysteri.txtGoth_05_5_1794_Radcl_TheMysteri.txt

    Goth_05_6_1794_Radcl_TheMysteri.txt

    Goth_05_7_1794_Radcl_TheMysteri.txt

    Goth_05_8_1794_Radcl_TheMysteri.txtGoth_05_9_1794_Radcl_TheMysteri.txt

    Goth_06_10_1796_Lewis_TheMonkARo.txtGoth_06_1_1796_Lewis_TheMonkARo.txtGoth_06_2_1796_Lewis_TheMonkARo.txt

    Goth_06_3_1796_Lewis_TheMonkARo.txt

    Goth_06_4_1796_Lewis_TheMonkARo.txt

    Goth_06_5_1796_Lewis_TheMonkARo.txtGoth_06_6_1796_Lewis_TheMonkARo.txt

    Goth_06_7_1796_Lewis_TheMonkARo.txt

    Goth_06_8_1796_Lewis_TheMonkARo.txt

    Goth_06_9_1796_Lewis_TheMonkARo.txtGoth_07_10_1797_Radcl_TheItalian.txtGoth_07_1_1797_Radcl_TheItalian.txtGoth_07_2_1797_Radcl_TheItalian.txt

    Goth_07_3_1797_Radcl_TheItalian.txt

    Goth_07_4_1797_Radcl_TheItalian.txt

    Goth_07_5_1797_Radcl_TheItalian.txt

    Goth_07_6_1797_Radcl_TheItalian.txtGoth_07_7_1797_Radcl_TheItalian.txt

    Goth_07_8_1797_Radcl_TheItalian.txt

    Goth_07_9_1797_Radcl_TheItalian.txt

    Goth_08_10_1799_Godwi_StLeonATal.txt

    Goth_08_1_1799_Godwi_StLeonATal.txt

    Goth_08_2_1799_Godwi_StLeonATal.txtGoth_08_3_1799_Godwi_StLeonATal.txt

    Goth_08_4_1799_Godwi_StLeonATal.txt

    Goth_08_5_1799_Godwi_StLeonATal.txt

    Goth_08_6_1799_Godwi_StLeonATal.txt

    Goth_08_7_1799_Godwi_StLeonATal.txt

    Goth_08_8_1799_Godwi_StLeonATal.txt

    Goth_08_9_1799_Godwi_StLeonATal.txt

    Goth_09_10_1806_Dacre_ZofloyaorT.txt

    Goth_09_1_1806_Dacre_ZofloyaorT.txt

    Goth_09_2_1806_Dacre_ZofloyaorT.txt

    Goth_09_3_1806_Dacre_ZofloyaorT.txt

    Goth_09_4_1806_Dacre_ZofloyaorT.txt

    Goth_09_5_1806_Dacre_ZofloyaorT.txt

    Goth_09_6_1806_Dacre_ZofloyaorT.txtGoth_09_7_1806_Dacre_ZofloyaorT.txt

    Goth_09_8_1806_Dacre_ZofloyaorT.txt

    Goth_09_9_1806_Dacre_ZofloyaorT.txt

    Goth_10_10_1810_Shell_ZastrozziA.txt

    h_10_1_1810_Shell_ZastrozziA.txt

    Goth_10_2_1810_Shell_ZastrozziA.txt

    Goth_10_3_1810_Shell_ZastrozziA.txt

    h_10_4_1810_Shell_ZastrozziA.txt

    Goth_10_5_1810_Shell_ZastrozziA.txt1810_Shell_ZastrozziA.txt

    Goth_10_7_1810_Shell_ZastrozziA.txt

    Goth_10_8_1810_Shell_ZastrozziA.txt

    _1810_Shell_ZastrozziA.txt

    Goth_11_10_1820_Matur_Melmoththe.txt

    Goth_11_1_1820_Matur_Melmoththe.txt

    Goth_11_2_1820_Matur_Melmoththe.txt

    Goth_11_3_1820_Matur_Melmoththe.txt

    Goth_11_4_1820_Matur_Melmoththe.txt

    Goth_11_5_1820_Matur_Melmoththe.txt

    Goth_11_6_1820_Matur_Melmoththe.txt

    Goth_11_7_1820_Matur_Melmoththe.txt

    Goth_11_8_1820_Matur_Melmoththe.txt

    Goth_11_9_1820_Matur_Melmoththe.txt

    Goth_12_10_1818_Shell_Frankenste.txtGoth_12_1_1818_Shell_Frankenste.txt

    Goth_12_2_1818_Shell_Frankenste.txt

    Goth_12_3_1818_Shell_Frankenste.txt

    Goth_12_4_1818_Shell_Frankenste.txt

    Goth_12_5_1818_Shell_Frankenste.txt

    Goth_12_6_1818_Shell_Frankenste.txt

    Goth_12_7_1818_Shell_Frankenste.txt

    Goth_12_8_1818_Shell_Frankenste.txtGoth_12_9_1818_Shell_Frankenste.txt

    Goth_13_10_1793_Smith_TheOldMano.txt

    Goth_13_1_1793_Smith_TheOldMano.txt

    Goth_13_2_1793_Smith_TheOldMano.txt

    Goth_13_3_1793_Smith_TheOldMano.txt

    Goth_13_4_1793_Smith_TheOldMano.txt

    Goth_13_5_1793_Smith_TheOldMano.txt

    Goth_13_6_1793_Smith_TheOldMano.txt

    Goth_13_7_1793_Smith_TheOldMano.txt

    Goth_13_8_1793_Smith_TheOldMano.txt

    Goth_13_9_1793_Smith_TheOldMano.txt

    Sens_01_10_1862_Bradd_LadyAudley.txt

    Sens_01_1_1862_Bradd_LadyAudley.txt

    Sens_01_2_1862_Bradd_LadyAudley.txt

    Sens_01_3_1862_Bradd_LadyAudley.txtSens_01_4_1862_Bradd_LadyAudley.txt

    Sens_01_5_1862_Bradd_LadyAudley.txt

    Sens_01_6_1862_Bradd_LadyAudley.txt

    Sens_01_7_1862_Bradd_LadyAudley.txt

    Sens_01_8_1862_Bradd_LadyAudley.txt

    Sens_01_9_1862_Bradd_LadyAudley.txt

    Sens_02_10_1868_Colli_TheMoonsto.txt

    Sens_02_1_1868_Colli_TheMoonsto.txt

    Sens_02_2_1868_Colli_TheMoonsto.txt

    Sens_02_3_1868_Colli_TheMoonsto.txt

    Sens_02_4_1868_Colli_TheMoonsto.txt

    Sens_02_5_1868_Colli_TheMoonsto.txt

    Sens_02_6_1868_Colli_TheMoonsto.txt

    Sens_02_7_1868_Colli_TheMoonsto.txt

    Sens_02_8_1868_Colli_TheMoonsto.txtSens_02_9_1868_Colli_TheMoonsto.txt

    Sens_03_10_1866_Colli_ArmadaleBy.txt

    Sens_03_1_1866_Colli_ArmadaleBy.txt

    Sens_03_2_1866_Colli_ArmadaleBy.txt

    Sens_03_3_1866_Colli_ArmadaleBy.txt

    Sens_03_4_1866_Colli_ArmadaleBy.txtSens_03_5_1866_Colli_ArmadaleBy.txt

    Sens_03_6_1866_Colli_ArmadaleBy.txt

    Sens_03_7_1866_Colli_ArmadaleBy.txt

    Sens_03_8_1866_Colli_ArmadaleBy.txt

    Sens_03_9_1866_Colli_ArmadaleBy.txt

    Sens_04_10_1859_Colli_TheWomanin.txt

    Sens_04_1_1859_Colli_TheWomanin.txt

    Sens_04_2_1859_Colli_TheWomanin.txt

    Sens_04_3_1859_Colli_TheWomanin.txtSens_04_4_1859_Colli_TheWomanin.txt

    Sens_04_5_1859_Colli_TheWomanin.txt

    Sens_04_6_1859_Colli_TheWomanin.txt

    Sens_04_7_1859_Colli_TheWomanin.txt

    Sens_04_8_1859_Colli_TheWomanin.txt

    Sens_04_9_1859_Colli_TheWomanin.txt

    Sens_05_10_1861_WoodH_EastLynneB.txt

    Sens_05_1_1861_WoodH_EastLynneB.txt

    Sens_05_2_1861_WoodH_EastLynneB.txt

    Sens_05_3_1861_WoodH_EastLynneB.txt

    Sens_05_4_1861_WoodH_EastLynneB.txtSens_05_5_1861_WoodH_EastLynneB.txtSens_05_6_1861_WoodH_EastLynneB.txtSens_05_7_1861_WoodH_EastLynneB.txt

    Sens_05_8_1861_WoodH_EastLynneB.txtSens_05_9_1861_WoodH_EastLynneB.txt

    Asides

    Citation

    CommunicatorRoles

    Comparing

    Constructive_Reasoning

    ContingentReasoning

    CuriosityRaising

    Decisiveness

    Defining

    Descriptive_Features

    DirectAddress

    Directives

    Excepting

    Exemplifying

    FirstPerson

    FollowingUp

    FormalQuery

    Future

    GeneralizingGeneric_firstPerson_Interior

    Give_Feedback

    Immediacy

    Intenseness

    Intimacy

    MetaDiscourse

    Narrative_Time

    Narrative_VP

    Negative_Emotion

    Negative_Relations

    NegativeValues

    OppositionalReasoning

    Past

    PersonRoles

    Positive_Relations

    PositiveValues

    Private_Cognitions

    Pronouns

    Public_Language Questions

    Quotations

    Reference_Abstract_Words

    Reference_Language

    Reporting_Change

    Reporting_Events

    Reporting_Process

    Reporting_States

    Requests

    Specifying

    Subjectivity

    Take_Responsibility

    Positive_Emotion

    5

    5

    0

    5

    PC1

    PC2

    0.3 0.2 0.1 0.0 0.1 0.2 0.3

    0.3

    0.2

    0.1

    0.0

    0.1

    0.2

    0.3

    0 5

    Figure 7.4: Docuscope scaterplo (ligh grey iles) and componen loadings (black).

    As we sudied our chars, i became clear ha hey resed on wo premises ha were

    quie dieren rom hose o curren genre heory: hey never looked a a genre per se, in

    isolaion, bu always and only in relaion o anoher genre; and hey were no ineresed

    in hose eaures ha could add up o a synheic ideal-ype, bu only in hose ha could

    difereniaeone genre rom an anoher. This relaional-dierenial emphasis made or a

    very realisic approach, reminiscen o Bourdieus posiion-aking: jus like auhors

    or schools, genres engage in a sruggle or recogniion: one could almos eel, no jus

    he dierence, bu he conico orms in hose rais ha pulled hem in one direcion or

    he oher. And ye, his image o genre was clearly also incomplee, because dierenial

    eaures may ell us all we need o know in order o demarcae one orm rom anoher,

    and ye very litle abou ha orms inner srucure. I all men in an audience wore pink,

    and all women blue, he colours would diereniae hem perecly, and ell us nohing

    abou hem. Well reurn o his poin a he end o he aricle.

  • 7/30/2019 Literary Lab Pamphlet 1

    21/29

    19

    GenresNational

    Silver Fork

    Two Genres in PCA Space

    PC1

    PC2

    6

    4

    2

    0

    4

    5 0 5

    2

    Figure 7.5: Mos Frequen Word scaterplo o wo genres raed on rs wo principal componens.

    GenresNational

    Silver Fork

    Two Genres in PCA Space

    PC1

    PC2

    6

    4

    2

    0

    4

    6 4 2 0 2 4

    2

    Figure 7.6: Docuscope scaterplo o wo genres raed on rs wo principal componens.

    Now, one hing ha he chars made clear was he variabiliyo genre signals: quie srong

    in gures 7.1-2, or insance, bu raher weak in abou one ourh o he cases like gures

    7.5-6, where neiher MFW nor Docuscope managed o exricae naional ales rom silver-

    ork novels. Why some genres should be so hard o separae especially in a case likehis, where he dierence, inuiively, ough o be quie vivid was an inriguing quesion;

    bu we decided o leave i or anoher sudy, and ocus insead on a group o chars where

  • 7/30/2019 Literary Lab Pamphlet 1

    22/29

    20

    he separaion was raher good, and dependen on a recurring se o rais: he pairings o

    gohic novels wih he hree ideological genres Jacobin, ani-Jacobin, and evangelical

    novels ha were heir shor-lived conemporaries.19 Since he chars were all similar, we

    reproduce here only he gohic/Jacobin pairings: gures. 7.7-8, based on MFW, and gures

    7.9-10, based on Docuscope and is Dimensions.

    Two Genres in PCA Space

    PC2

    4

    2

    0

    4

    2

    PC1

    5 50

    GenresGothic

    Jacobin

    Figure 7.7: Mos Frequen Word scaterplo o wo genres raed on rs wo principal componens.

    To beter undersand he relaionship beween he wo genresand o begin o pu he

    gures ino languagewe looked closely a he eaures ha were paricularly eecive

    a separaing gohic rom Jacobin along he rs principal componen (PC1: he x-axis in

    gures 7.7-10). A principal componen ranks he likelihood o cerain eaures occurring,

    so exs are sored according o he eaures hey lack, as well as by he eaures hey have.

    19 One o our problems was ha we had auomaed our comparisons, using only he rs wo (and hereore, mos

    powerul) principal componens o pull apar he genres. O course, PCA generaes muliple componens and here

    are ways o esablishing (or example, he Tukey es) wheher any given componen sors wo groups. Bu wewaned some raw measure o sorabiliy among pairs, which is wha led us o simply prole all o he pairs on heir

    rs wo componens and leave oher poenially quie powerul componens aside.

  • 7/30/2019 Literary Lab Pamphlet 1

    23/29

    21

    5

    0

    5

    PC1

    PC2

    0.3 0.2 0.1 0.0 0.1 0.2 0.3

    5 0 5

    Goth_01_1_1786_Beckf_VathekTran

    Goth_01_10_1786_Beckf_VathekTran

    Goth_01_2_1786_Beckf_VathekTran

    Goth_01_3_1786_Beckf_VathekTran

    Goth_01_4_1786_Beckf_VathekTran

    Goth_01_5_1786_Beckf_VathekTran

    Goth_01_6_1786_Beckf_VathekTran

    Goth_01_7_1786_Beckf_VathekTran

    Goth_01_8_1786_Beckf_VathekTran

    Goth_01_9_1786_Beckf_VathekTran

    Goth_02_1_1788_Smith_Emmelineth

    Goth_02_10_1788_Smith_Emmelineth

    Goth_02_2_1788_Smith_Emmelineth

    Goth_02_3_1788_Smith_EmmelinethGoth_02_4_1788_Smith_Emmelineth

    Goth_02_5_1788_Smith_Emmelineth

    Goth_02_6_1788_Smith_EmmelinethGoth_02_7_1788_Smith_Emmelineth Goth_02_8_1788_Smith_Emmelineth

    Goth_02_9_1788_Smith_Emmelineth

    th_03_1_1790_Radcl_ASicilianR

    Goth_03_10_1790_Radcl_ASicilianR

    Goth_03_2_1790_Radcl_ASicilianR

    Goth_03_3_1790_Radcl_ASicilianR

    Goth_03_4_1790_Radcl_ASicilianR_5_1790_Radcl_ASicilianR

    Goth_03_6_1790_Radcl_ASicilianR

    Goth_03_7_1790_Radcl_ASicilianRGoth_03_8_1790_Radcl_ASicilianR

    Goth_03_9_1790_Radcl_ASicilianR

    Goth_04_1_1791_Radcl_TheRomance

    Goth_04_10_1791_Radcl_TheRomance

    Goth_04_2_1791_Radcl_TheRomance

    Goth_04_3_1791_Radcl_TheRomanceGoth_04_4_1791_Radcl_TheRomance

    Goth_04_5_1791_Radcl_TheRomance

    Goth_04_6_1791_Radcl_TheRomance

    Goth_04_7_1791_Radcl_TheRomance

    Goth_04_8_1791_Radcl_TheRomance

    Goth_04_9_1791_Radcl_TheRomance

    Goth_05_1_1794_Radcl_TheMysteri

    Goth_05_10_1794_Radcl_TheMysteriGoth_05_2_1794_Radcl_TheMysteri

    Goth_05_3_1794_Radcl_TheMysteri

    Goth_05_4_1794_Radcl_TheMysteri

    Goth_05_5_1794_Radcl_TheMysteri

    Goth_05_6_1794_Radcl_TheMysteri

    Goth_05_7_1794_Radcl_TheMysteri

    Goth_05_8_1794_Radcl_TheMysteri

    Goth_05_9_1794_Radcl_TheMysteri

    Goth_06_1_1796_Lewis_TheMonkARo

    Goth_06_10_1796_Lewis_TheMonkARo

    Goth_06_2_1796_Lewis_TheMonkARoGoth_06_3_1796_Lewis_TheMonkARo

    Goth_06_4_1796_Lewis_TheMonkARo

    Goth_06_5_1796_Lewis_TheMonkARo

    Goth_06_6_1796_Lewis_TheMonkARo

    Goth_06_7_1796_Lewis_TheMonkARo

    Goth_06_8_1796_Lewis_TheMonkARo

    Goth_06_9_1796_Lewis_TheMonkARo

    Goth_07_1_1797_Radcl_TheItalian

    Goth_07_10_1797_Radcl_TheItalian

    Goth_07_2_1797_Radcl_TheItalian

    Goth_07_3_1797_Radcl_TheItalianGoth_07_4_1797_Radcl_TheItalian

    Goth_07_5_1797_Radcl_TheItalian

    Goth_07_6_1797_Radcl_TheItalian

    Goth_07_7_1797_Radcl_TheItalian

    Goth_07_8_1797_Radcl_TheItalian

    Goth_07_9_1797_Radcl_TheItalian

    Goth_08_1_1799_Godwi_StLeonATal

    Goth_08_10_1799_Godwi_StLeonATal

    Goth_08_2_1799_Godwi_StLeonATal

    Goth_08_3_1799_Godwi_StLeonATal

    Goth_08_4_1799_Godwi_StLeonATalGoth_08_5_1799_Godwi_StLeonATal

    Goth_08_6_1799_Godwi_StLeonATal

    Goth_08_7_1799_Godwi_StLeonATal

    Goth_08_8_1799_Godwi_StLeonATal

    Goth_08_9_1799_Godwi_StLeonATal

    Goth_09_1_1806_Dacre_ZofloyaorT

    Goth_09_10_1806_Dacre_ZofloyaorT

    Goth_09_2_1806_Dacre_ZofloyaorT

    Goth_09_3_1806_Dacre_ZofloyaorT

    Goth_09_4_1806_Dacre_ZofloyaorT

    Goth_09_5_1806_Dacre_ZofloyaorT

    Goth_09_6_1806_Dacre_ZofloyaorT

    Goth_09_7_1806_Dacre_ZofloyaorT

    Goth_09_8_1806_Dacre_ZofloyaorT

    Goth_09_9_1806_Dacre_ZofloyaorT

    Goth_10_1_1810_Shell_ZastrozziA

    Goth_10_10_1810_Shell_ZastrozziA

    Goth_10_2_1810_Shell_ZastrozziA

    Goth_10_3_1810_Shell_ZastrozziA

    Goth_10_4_1810_Shell_ZastrozziA

    Goth_10_5_1810_Shell_ZastrozziA

    Goth_10_6_1810_Shell_ZastrozziA

    Goth_10_7_1810_Shell_ZastrozziA

    Goth_10_8_1810_Shell_ZastrozziA

    Goth_10_9_1810_Shell_ZastrozziA

    Goth_11_1_1820_Matur_Melmoththe

    Goth_11_10_1820_Matur_Melmoththe

    Goth_11_2_1820_Matur_Melmoththe

    Goth_11_3_1820_Matur_Melmoththe

    Goth_11_4_1820_Matur_Melmoththe

    Goth_11_5_1820_Matur_MelmoththeGoth_11_6_1820_Matur_Melmoththe

    Goth_11_7_1820_Matur_Melmoththe

    Goth_11_8_1820_Matur_Melmoththe

    Goth_11_9_1820_Matur_Melmoththe

    Goth_12_1_1818_Shell_Frankenste

    Goth_12_10_1818_Shell_Frankenste

    Goth_12_2_1818_Shell_Frankenste

    Goth_12_3_1818_Shell_Frankenste

    Goth_12_4_1818_Shell_Frankenste

    Goth_12_5_1818_Shell_Frankenste

    Goth_12_6_1818_Shell_Frankenste

    Goth_12_7_1818_Shell_FrankensteGoth_12_8_1818_Shell_FrankensteGoth_12_9_1818_Shell_Frankenste

    Goth_13_1_1793_Smith_TheOldMano

    Goth_13_10_1793_Smith_TheOldMano

    Goth_13_2_1793_Smith_TheOldMano

    Goth_13_3_1793_Smith_TheOldMano

    Goth_13_4_1793_Smith_TheOldMano

    Goth_13_5_1793_Smith_TheOldManoGoth_13_6_1793_Smith_TheOldMano

    Goth_13_7_1793_Smith_TheOldMano

    Goth_13_8_1793_Smith_TheOldMano

    Goth_13_9_1793_Smith_TheOldMano

    Jaco_01_1_1796_BageR_Hermsprong

    Jaco_01_10_1796_BageR_HermsprongJaco_01_2_1796_BageR_Hermsprong

    Jaco_01_3_1796_BageR_Hermsprong

    Jaco_01_4_1796_BageR_Hermsprong

    Jaco_01_5_1796_BageR_Hermsprong

    Jaco_01_6_1796_BageR_H

    Jaco_01_7_1796_BageR_Hermsprong

    Jaco_01_8_1796_BageR_Hermsprong

    Jaco_01_9_1796_BageR_Hermsprong

    Jaco_02_1_1794_Godwi_ThingsAsTh

    Jaco_02_10_1794_Godwi_ThingsAsTh

    Jaco_02_2_1794_Godwi_ThingsAsTh

    Jaco_02_3_1794_Godwi_ThingsAsTh

    Jaco_02_4_1794_Godwi_ThingsAsTh

    Jaco_02_5_1794_Godwi_ThingsAsTh

    Jaco_02_6_1794_Godwi_ThingsAsTh

    Jaco_02_7_1794_Godwi_ThingsAsThJaco_02_8_1794_Godwi_ThingsAsTh

    Jaco_02_9_1794_Godwi_ThingsAsTh

    Jaco_03_1_1796_HaysM_MemoirsofE

    Jaco_03_10_1796_HaysM_MemoirsofE

    Jaco_03_2_1796_HaysM_MemoirsofE

    Jaco_03_3_1796_HaysM_MemoirsofE

    Jaco_03_4_1796_HaysM_MemoirsofE

    Jaco_03_5_1796_HaysM_MemoirsofE

    Jaco_03_6_1796_HaysM_MemoirsofE

    Jaco_03_7_1796_HaysM_MemoirsofEJaco_03_8_1796_HaysM_MemoirsofE

    Jaco_03_9_1796_HaysM_MemoirsofE

    Jaco_04_1_1792_Holcr_AnnaS

    Jaco_04_10_1792_Holcr_AnnaStIves

    Jaco_04_2_1792_Holcr_AnnaStIves

    Jaco_04_3_1792_Holcr_AnnaStIves

    Jaco_04_4_1792_Holcr_Ann

    Jaco_04_5_1792_Holcr_AnnaStIves

    Jaco_04_6_1792_Holcr_Ann

    Jaco_04_7_1792_Holcr_AnnaStIves

    Jaco_04_8_1792_Holcr_AnnaStIvesJaco_04_9_1792_Holcr_AnnaStIves

    Jaco_05_1_1794_Holcr_TheAdventuJaco_05_10_1794_Holcr_TheAdventuJaco_05_2_1794_Holcr_TheAdventu

    Jaco_05_3_1794_Holcr_TheAdventu

    Jaco_05_4_1794_Holcr_TheAdventu

    Jaco_05_5_1794_Holcr_TheAdventu

    Jaco_05_6_1794_Holcr_TheAdventuJaco_05_7_1794_Holcr_TheAdventu

    Jaco_05_8_1794_Holcr_TheAdventu

    Jaco_05_9_1794_Holcr_TheAdventu

    Jaco_06_1_1796_Inchb_NatureandA

    Jaco_06_10_1796_Inchb_NatureandA

    Jaco_06_2_1796_Inchb_NatureandA

    Jaco_06_3_1796_Inchb_NatureandA

    Jaco_06_4_1796_Inchb_NatureandA

    Jaco_06_5_1796_Inchb_NatureandA

    Jaco_06_6_1796_Inchb_NatureandA

    Jaco_06_7_1796_Inchb_NatureandA

    Jaco_06_8_1796_Inchb_NatureandA

    Jaco_06_9_1796_Inchb_NatureandA

    Jaco_07_1_1791_Inchb_ASimpleSto

    Jaco_07_10_1791_Inchb_ASimpleSto

    Jaco_07_2_1791_Inchb_ASimpleSto

    Jaco_07_3_1791_Inchb_ASimpleSto

    Jaco_07_4_1791_Inchb_ASimpleSto

    Jaco_07_5_1791_Inchb_ASimpleSto

    Jaco_07_6_1791_Inchb_ASimpleSto

    Jaco_07_7_1791_Inchb_ASimpleStoJaco_07_8_1791_Inchb_ASimpleStoJaco_07_9_1791_Inchb_ASimpleSto

    Jaco_08_1_1799_HaysM_TheVictimo

    Jaco_08_10_1799_HaysM_TheVictimo

    Jaco_08_2_1799_HaysM_TheVictimo

    Jaco_08_3_1799_HaysM_TheVictimo

    Jaco_08_4_1799_HaysM_TheVictimo

    Jaco_08_5_1799_HaysM_TheVictimo

    Jaco_08_6_1799_HaysM_TheVictimo

    Jaco_08_7_1799_HaysM_TheVictimo

    Jaco_08_8_1799_HaysM_TheVictimo

    Jaco_08_9_1799_HaysM_TheVictimo

    Jaco_09_1_1798_Wolls_TheWrongso

    Jaco_09_10_1798_Wolls_TheWrongso

    Jaco_09_2_1798_Wolls_TheWrongso

    Jaco_09_3_1798_Wolls_TheWrongso

    Jaco_09_4_1798_Wolls_TheWrongso

    Jaco_09_5_1798_Wolls_TheWrongso

    Jaco_09_6_1798_Wolls_TheWrongso

    Jaco_09_7_1798_Wolls_TheWrongso

    Jaco_09_8_1798_Wolls_TheWrongso

    Jaco_09_9_1798_Wolls_TheWrongso

    a

    all

    an

    and

    as

    at

    be

    but

    by

    for

    from

    had

    have

    heher

    him

    his

    i

    if

    in

    is

    it

    me

    my

    nonot

    of

    on

    or

    p_apos

    p_comma

    p_exlam

    p_hyphen

    p_period

    p_ques

    p_quote

    p_semi

    said

    she

    so

    thatthe

    they

    this

    to

    was

    were

    what

    when

    which

    who

    will

    with

    would

    you

    your

    0.3

    0.2

    0.1

    0.0

    0.1

    0.2

    0.3

    Figure 7.8: Mos Frequen Word scaterplo wih iles (ligh grey) and componen loadings (black).

    Roughly speaking, we ound ha he gohic novel averages less alk and more acion han

    he Jacobin. Words and phrases ha characerize gohic exs show a marked narraive

    inclinaion: pas ense and pronouns; spaial preposiions; and words marked by Docus-

    cope as Narraive Time (or example, whils, when he, as he). See in MFW, on he

    le side o gure 7.8: was, had, who, she, he, her, his, hey, he ubiquious he, and he

    large group o locaive preposiions rom, on, in, a; in Docuscope, on he le side o g-

    ure 7.10, see Narraive VP (or example, heard he, reached he, commanded he) and

    Pronouns. Markers associaed wih oral discourse, on he oher hand, end no o occur in

    gohic novels: a oregrounding o he addressee (you, your), quesions, polemical markers

    (bu, no), and verbs ineced in he presen, uure and condiional. In MFW, noe on he

    righhand side o gure 7.8 he cluser o you, p_ques [?], bu, i, no, is, will, and would;

    in Docuscope, on he righ o gure 7.10, see Quesions, Opposiional Reasoning (no,

    bu, however), and, jus below i, Direcives (should, mus, you will soon). Though

    Jacobin exs generally ack in his direcion, hey are more scatered han he gohic ones.Wha separaes he genres here seems o be, no so much he absence o narraive in

    Jacobin exs, bu he presence o alk, somehing like an argumenaive syle.

  • 7/30/2019 Literary Lab Pamphlet 1

    24/29

    22

    Two Genres in PCA Space

    PC2

    6

    4

    2

    0

    2

    6

    PC1

    4 26 0 2 4 6

    GenresGothic

    Jacobin

    Figure 7.9: Docuscope scaterplo o wo genres raed on rs wo principal componens.

  • 7/30/2019 Literary Lab Pamphlet 1

    25/29

    23

    5

    0

    5

    PC1

    PC2

    0.3 0.2 0.1 0.0 0.1 0.2 0.3

    0.3

    0.2

    0.1

    0.0

    0.1

    0

    .2

    0.3

    5 0 5

    Goth_01_10_1786_Beckf_VathekTran.txt

    Goth_01_1_1786_Beckf_VathekTran.txt

    Goth_01_2_1786_Beckf_VathekTran.txtGoth_01_3_1786_Beckf_VathekTran.txt

    Goth_01_4_1786_Beckf_VathekTran.txtGoth_01_5_1786_Beckf_VathekTran.txt

    Goth_01_6_1786_Beckf_VathekTran.txt

    Goth_01_7_1786_Beckf_VathekTran.txt

    Goth_01_8_1786_Beckf_VathekTran.txt

    Goth_01_9_1786_Beckf_VathekTran.txt

    Goth_02_10_1788_Smith_Emmelineth.txt

    Goth_02_1_1788_Smith_Emmelineth.txt

    Goth_02_2_1788_Smith_Emmelineth.txt

    Goth_02_3_1788_Smith_Emmelineth.txt

    Goth_02_4_1788_Smith_Emmelineth.txtGoth_02_5_1788_Smith_Emmelineth.txt

    Goth_02_6_1788_Smith_Emmelineth.txt

    Goth_02_7_1788_Smith_Emmelineth.txt

    Goth_02_8_1788_Smith_Emmelineth.txtGoth_02_9_1788_Smith_Emmelineth.txt

    Goth_03_10_1790_Radcl_ASicilianR.txt

    Goth_03_1_1790_Radcl_ASicilianR.txt

    Goth_03_2_1790_Radcl_ASicilianR.txt

    Goth_03_3_1790_Radcl_ASicilianR.txt

    Goth_03_4_1790_Radcl_ASicilianR.txt

    Goth_03_5_1790_Radcl_ASicilianR.txt

    Goth_03_6_1790_Radcl_ASicilianR.txt

    Goth_03_7_1790_Radcl_ASicilianR.txt

    Goth_03_8_1790_Radcl_ASicilianR.txtGoth_03_9_1790_Radcl_ASicilianR.txt

    Goth_04_10_1791_Radcl_TheRomance.txt

    Goth_04_1_1791_Radcl_TheRomance.txt

    Goth_04_2_1791_Radcl_TheRomance.txt

    Goth_04_3_1791_Radcl_TheRomance.txt

    Goth_04_4_1791_Radcl_TheRomance.txt

    Goth_04_5_1791_Radcl_TheRomance.txt

    Goth_04_6_1791_Radcl_TheRomance.txt

    Goth_04_7_1791_Radcl_TheRomance.txt

    Goth_04_8_1791_Radcl_TheRomance.txt

    Goth_04_9_1791_Radcl_TheRomance.txt

    Goth_05_10_1794_Radcl_TheMysteri.txt

    Goth_05_1_1794_Radcl_TheMysteri.txt

    Goth_05_2_1794_Radcl_TheMysteri.txt

    Goth_05_3_1794_Radcl_TheMysteri.txt

    Goth_05_4_1794_Radcl_TheMysteri.txtGoth_05_5_1794_Radcl_TheMysteri.txtGoth_05_6_1794_Radcl_TheMysteri.txt

    Goth_05_7_1794_Radcl_TheMysteri.txt

    Goth_05_8_1794_Radcl_TheMysteri.txtGoth_05_9_1794_Radcl_TheMysteri.txt

    Goth_06_10_1796_Lewis_TheMonkARo.txt

    Goth_06_1_1796_Lewis_TheMonkARo.txt

    Goth_06_2_1796_Lewis_TheMonkARo.txt

    Goth_06_3_1796_Lewis_TheMonkARo.txt

    Goth_06_4_1796_Lewis_TheMonkARo.txt

    Goth_06_5_1796_Lewis_TheMonkARo.txtGoth_06_6_1796_Lewis_TheMonkARo.txt

    Goth_06_7_1796_Lewis_TheMonkARo.txtGoth_06_8_1796_Lewis_TheMonkARo.txt

    Goth_06_9_1796_Lewis_TheMonkARo.txt

    Goth_07_10_1797_Radcl_TheItalian.txt

    Goth_07_1_1797_Radcl_TheItalian.txt

    Goth_07_2_1797_Radcl_TheItalian.txt

    Goth_07_3_1797_Radcl_TheItalian.txtGoth_07_4_1797_Radcl_TheItalian.txt

    Goth_07_5_1797_Radcl_TheItalian.txtGoth_07_6_1797_Radcl_TheItalian.txt

    Goth_07_7_1797_Radcl_TheItalian.txt

    Goth_07_8_1797_Radcl_TheItalian.txt

    Goth_07_9_1797_Radcl_TheItalian.txt

    Goth_08_10_1799_Godwi_StLeonATal.txt

    Goth_08_1_1799_Godwi_StLeonATal.txt

    Goth_08_2_1799_Godwi_StLeonATal.txt

    Goth_08_3_1799_Godwi_StLeonATal.txt

    Goth_08_4_1799_Godwi_StLeonATal.txt

    Goth_08_5_1799_Godwi_StLeonATalGoth_08_6_1799_Godwi_StLeonATal.txt

    Goth_08_7_1799_Godwi_StLeonATal.txtGoth_08_8_1799_Godwi_StLeonATal.txt

    Goth_08_9_1799_Godwi_StLeonATal.txt

    Goth_09_10_1806_Dacre_ZofloyaorT.txt

    Goth_09_1_1806_Dacre_ZofloyaorT.txt

    Goth_09_2_1806_Dacre_ZofloyaorT.txtGoth_09_3_1806_Dacre_ZofloyaorT.txt

    Goth_09_4_1806_Dacre_ZofloyaorT.txt

    Goth_09_5_1806_Dacre_ZofloyaorT.txt

    Goth_09_6_1806_Dacre_ZofloyaorT.txt

    Goth_09_7_1806_Dacre_ZofloyaorT.txtGoth_09_8_1806_Dacre_ZofloyaorT.txt

    Goth_09_9_1806_Dacre_ZofloyaorT.txt

    Goth_10_10_1810_Shell_ZastrozziA.txt

    Goth_10_1_1810_Shell_ZastrozziA.txt

    Goth_10_2_1810_Shell_ZastrozziA.txt

    Goth_10_3_1810_Shell_ZastrozziA.txt

    Goth_10_4_1810_Shell_ZastrozziA.txt

    Goth_10_5_1810_Shell_ZastrozziA.txt

    Goth_10_6_1810_Shell_ZastrozziA.txt

    Goth_10_7_1810_Shell_ZastrozziA.txt

    Goth_10_8_1810_Shell_ZastrozziA.txt

    Goth_10_9_1810_Shell_ZastrozziA.txt

    Goth_11_10_1820_Matur_Melmoththe.txt

    Goth_11_1_1820_Matur_Melmoththe.txt

    Goth_11_2_1820_Matur_Melmoththe.txt

    Goth_11_3_1820_Matur_Melmoththe.txt

    Goth_11_4_1820_Matur_Melmoththe.txt

    Goth_11_5_1820_Matur_Melmoththe.txtGoth_11_6_1820_Matur_Melmoththe.txt

    Goth_11_7_1820_Matur_Melmoththe.txtGoth_11_8_1820_Matur_Melmoththe.txt

    Goth_11_9_1820_Matur_Melmoththe.txt

    Goth_12_10_1818_Shell_Frankenste.txt

    Goth_12_1_1818_Shell_Frankenste.txt

    Goth_12_2_1818_Shell_Frankenste.txt

    Goth_12_3_1818_Shell_Frankenste.txt

    Goth_12_4_1818_Shell_Frankenste.txt

    Goth_12_5_1818_Shell_Frankenste.txt

    Goth_12_6_1818_Shell_Frankenste.txt

    Goth_12_7_1818_Shell_Frankenste.txt

    Goth_12_8_1818_Shell_Frankenste.txt

    Goth_12_9_1818_Shell_Frankenste.txt

    Goth_13_10_1793_Smith_TheOldMano.txt

    Goth_13_1_1793_Smith_TheOldMano.txt

    Goth_13_2_1793_Smith_TheOldMano.txtGoth_13_3_1793_Smith_TheOldMano.txt

    Goth_13_4_1793_Smith_TheOldMano.txt

    Goth_13_5_1793_Smith_TheOldMano.txt

    Goth_13_6_1793_Smith_TheOldMano.txt

    Goth_13_7_1793_Smith_TheOldMano.txt

    Goth_13_8_1793_Smith_TheOldMano.txt

    Goth_13_9_1793_Smith_TheOldMano.txt

    Jaco_01_10_1796_BageR_Hermsprong.txtJaco_01_1_1796_BageR_Hermsprong.txt

    Jaco_01_2_1796_BageR_Hermsprong.txt

    Jaco_01_3_1796_BageR_Hermsprong.txtJaco_01_4_1796_BageR_Hermsprong.txt

    Jaco_01_5_1796_BageR_Hermsprong.txtJaco_01_6_1796_BageR_Hermsprong.txt

    Jaco_01_7_1796_BageR_Hermsprong.txt

    Jaco_01_8_1796_BageR_Hermsprong.txt

    Jaco_01_9_1796_BageR_Hermsprong.txt

    Jaco_02_10_1794_Godwi_ThingsAsTh.txt

    Jaco_02_1_1794_Godwi_ThingsAsTh.txt

    Jaco_02_2_1794_Godwi_ThingsAsTh.txt

    Jaco_02_3_1794_Godwi_ThingsAsTh.txt

    Jaco_02_4_1794_Godwi_ThingsAsTh.txtJaco_02_5_1794_Godwi_ThingsAsTh.txt

    Jaco_02_6_1794_Godwi_ThingsAsTh.txt

    Jaco_02_7_1794_Godwi_ThingsAsTh.txtJaco_02_8_1794_Godwi_ThingsAsTh.txt

    Jaco_02_9_1794_Godwi_ThingsAsTh.txt

    Jaco_03_10_1796_HaysM_MemoirsofE.txt

    Jaco_03_1_1796_HaysM_MemoirsofE.txtJaco_03_2_1796_HaysM_MemoirsofE.txt

    Jaco_03_3_1796_HaysM_MemoirsofE.txt

    Jaco_03_4_1796_HaysM_MemoirsofE.txt

    Jaco_03_5_1796_HaysM_MemoirsofE.txt

    Jaco_03_6_1796_HaysM_MemoirsofE.txt

    Jaco_03_7_1796_HaysM_MemoirsofE.txt

    Jaco_03_8_1796_HaysM_MemoirsofE.txt

    Jaco_03_9_1796_HaysM_MemoirsofE.txt

    Jaco_04_10_1792_Holcr_AnnaStIves.txt Jaco_04_1_1792_Holcr_AnnaStIv

    Jaco_04_2_1792_Holcr_AnnaStIves.txt

    Jaco_04_3_1792_Holcr_AnnaStIves

    Jaco_04_4_1792_Holcr_A

    Jaco_04_5_1792_Holcr_AnnaStIves.txt

    Jaco_04_6_1792_Holcr_Ann

    Jaco_04_7_1792_Holcr_AnnaStIv

    Jaco_04_8_1792_Holcr_AnnaStIves.txt

    Jaco_04_9_1792_Holcr_AnnaStIves.txt

    Jaco_05_10_1794_Holcr_TheAdventu.txt

    Jaco_05_1_1794_Holcr_TheAdventu.txtJaco_05_2_1794_Holcr_TheAdventu.txt

    Jaco_05_3_1794_Holcr_TheAdventu.txt

    Jaco_05_4_1794_Holcr_TheAdventu.txt

    Jaco_05_5_1794_Holcr_TheAdventu.txt

    Jaco_05_6_1794_Holcr_TheAdventu.txtJaco_05_7_1794_Holcr_TheAdventu.txt

    Jaco_05_8_1794_Holcr_TheAdventu.txt

    Jaco_05_9_1794_Holcr_TheAdventu.txt

    Jaco_06_10_1796_Inchb_NatureandA.txt

    Jaco_06_1_1796_Inchb_NatureandA.txt

    Jaco_06_2_1796_Inchb_NatureandA.txtJaco_06_3_1796_Inchb_NatureandA.txt

    Jaco_06_4_1796_Inchb_NatureandA.txt

    Jaco_06_5_1796_Inchb_NatureandA.txt

    Jaco_06_6_1796_Inchb_NatureandA.txt

    Jaco_06_7_1796_Inchb_NatureandA.txt

    Jaco_06_8_1796_Inchb_NatureandA.txt

    Jaco_06_9_1796_Inchb_NatureandA.txt

    Jaco_07_10_1791_Inchb_ASimpleSto.txt

    Jaco_07_1_1791_Inchb_ASimpleSto.txtJaco_07_2_1791_Inchb_ASimpleSto.txt

    Jaco_07_3_1791_Inchb_ASimpleSto.txt

    Jaco_07_4_1791_Inchb_ASimpleSto.txt

    Jaco_07_5_1791_Inchb_ASimpleSto.txt

    Jaco_07_6_1791_Inchb_ASimpleSto.txt

    Jaco_07_7_1791_Inchb_ASimpleSto.txt

    Jaco_07_8_1791_Inchb_ASimpleSto.txt

    Jaco_07_9_1791_Inchb_ASimpleSto.txt

    Jaco_08_10_1799_HaysM_TheVictimo.txt

    Jaco_08_1_1799_HaysM_TheVictimo.txt

    Jaco_08_2_1799_HaysM_TheVictimo.txt

    Jaco_08_3_1799_HaysM_TheVictimo.txt

    Jaco_08_4_1799_HaysM_TheVictimo.txt

    Jaco_08_5_1799_HaysM_TheVictimo.txt

    Jaco_08_6_1799_HaysM_TheVictimo.txt

    Jaco_08_7_1799_HaysM_TheVictimo.txt

    Jaco_08_8_1799_HaysM_TheVictimo.txt

    Jaco_08_9_1799_HaysM_TheVictimo.txt

    Jaco_09_10_1798_Wolls_TheWrongso.txt

    Jaco_09_1_1798_Wolls_TheWrongso.txt

    Jaco_09_2_1798_Wolls_TheWrongso.txt

    Jaco_09_3_1798_Wolls_TheWrongso.txt

    Jaco_09_4_1798_Wolls_TheWrongso.txtJaco_09_5_1798_Wolls_TheWrongso.txt

    Jaco_09_6_1798_Wolls_TheWrongso.txt

    Jaco_09_7_1798_Wolls_TheWrongso.txtJaco_09_8_1798_Wolls_TheWrongso.txt

    Jaco_09_9_1798_Wolls_TheWrongso.txt

    Asides

    Citation

    CommunicatorRoles

    Comparing

    ConstructiveReasoning

    ContingentReasoning

    CuriosityRaising

    Decisiveness

    Defining

    scriptive_Features

    DirectAddress

    Directives

    Excepting

    Exemplifying

    FirstPerson_Interior

    F ollowing_Up

    Formal_Query

    FutureGeneralizing

    Generic_First_Person

    GiveFeedback

    Immediacy

    Intenseness

    Intimacy

    MetaDiscourse

    Narrative_Timearrative_VP

    Negative_Emotion

    Negative_Relations

    NegativeValues

    OppositionalReasoning

    Past

    PersonRoles

    P ositive_Emotion

    Positive_Relations

    Positive_Values

    Private_Cognitions

    Pronouns Public_Language

    Questions

    Quotations

    Reference_Abstract_Words

    Reference_Language

    Reporting_Change

    Reporting_Ev

    Figure 7.10: Docuscope scaterplo wih iles (ligh grey) and componen loadings (black).

    Such, hen, were he raw daa ha our analyic echniques had placed in ron o us.

    Could hey become good inerpreive quesions? We ried. Noicing, or insance, he

    high requency o he condiional in he ideological genres where, indeed, possibiliy

    is imporan Jockers and Moreti compiled a lis o he (more or less) 13,000 senences

    ha included would; looked a he associaed pronouns, adjecives, and adverbs; a he

    ypes o verbs involved; a he negaive orms, he pas ense ... A ew resuls sood ou:

    would+never occurred wice as oen in he gloomy evangelical novels han elsewhere,

    or insance; and he impersonal pronoun i was 50% more requen in Jacobin and ani-

    Jacobin novels ull o absrac discussions o principle han anywhere else. Boh nd-

    ings made perec sense. Bu were hey also surprising? They cerainly corroboraed and

    enriched exising knowledge o he genres in quesion. Did hey also changei?

  • 7/30/2019 Literary Lab Pamphlet 1

    26/29

    24

    8. March 2010: Experimens, Exploraions, Hypoheses

    In March, we me or one las rerospecive glance a a year o work. Why had we urned o

    Docuscope and MFW in he rs place? Because we were looking or an explici, quani-

    able way o assign exs o his or ha genre. I was, in par a leas, a mater o atribuion.

    Atribuion ... To race every piece o is real creaor, wries Carlo Ginzburg,

    we should no depend () on he mos conspicuous characerisics o a pain-

    ing, which are he easies o imiae: eyes raised owards he heavens in he

    gures o Perugino, Leonardos smiles, and so on. We should examine, insead,

    he mos rivial deails ha would have been inuenced leas by he manner-

    isms o he ariss school: earlobes, ngernails, shapes o ngers and o oes.

    Earlobes, ngernails ... I is in hese involunary signs, Ginzburg coninues,

    in he maerial ries a calligrapher migh call hem ourishes compa-

    rable o avorie words and phrases which mos people inroduce ino heir

    speaking and wriing uninenionally, oen wihou realizing i, ha Morelli

    recognized he sures clue o an ariss ideniy.20

    Involunary signs: his is ceranly wha MFW and LATs are. Bu are hey jusha? Because,

    clearly, here is a problem wih earlobes and ngernails: good as hey migh be a ideniy-

    ing he auhor o a paining, hey are worhless a explaining is meaning. In ac, hey are

    good a he one becausehey are bad a he oher: is only because ries have no sruc-

    ural uncion, ha auhors le go and wrie uninenionally, wihou realizing i hereby

    beraying hemselves. I hose words were imporan, hey would be more careul.

    There is somehing paradoxical in hese rais ha classiy so well, and explain so litle.Especially so in our case: because, aer all, MFW and LATs were in a leas one respec

    he very opposie o earlobes and ngernails: insead o being rare and peripheral deails,

    hey were so requen as o be almos ubiquious. And how could such pervasive rais ell

    us nohing abou he srucure o genre? I was possible, o course, ha i was all our aul;

    ha, alhough we had managed o isolae he daa, and were probably he rs o see

    hem, we jus didn know how o make sense o hem. Possible; and we are ready o place

    our daa a he disposal o ohers, who may obain beter resuls.

    Bu here is also a simpler explanaion: namely, ha hese eaures which are so eec-

    ive a diereniaing genres, and so enwined wih heir overall exure hese eaurescanno oer new insighs ino srucure, because hey aren independen rais, bu mere

    consequences o higher-order choices. Do you wan o wrie a sory where each and every

    room may be ull o surprises? Then locaive preposiions, aricles and verbs in he pas

    ense are bound o ollow. They are he efecso he chosen narraive srucure. And, yes,

    once Docuscope and MFW oreground hem, making us ully aware o heir presence, our

    knowledge is analyically enriched: we see he space o he gohic, or he link beween

    acion verbs and objecs (highlighed by he requency o aricles), wih much greaer

    clariy. Bu, or he ime being, he gain seems o be comparaive more han qualiaive:

    greaerclariy, raher han clariy o a dieren ype.

    20 Carlo Ginzburg, Clues, in Clues, Myths, and the Historical Method, Hopkins UP 1989, pp. 96-7, 118.

  • 7/30/2019 Literary Lab Pamphlet 1

    27/29

    25

    We sared wih an experimen: esing he classiying power o Docuscope in a new and

    conrolled seting. The experimen hen urned ino an exploraion: Docuscope and MFW,

    charing he eld o novelisic genres, and heir inner composiion. Exploraory Daa

    Analysis, as John Tukey has called i: deecive work, ocusing on clues ha lead o new

    quesions, and a broader undersanding o he daa. Saisical ndings, said Heuser, made

    us realize ha genres are icebergs: wih a visible porion oaing above he waer, and amuch larger par hidden below, and exending o unknown dephs. Realizing ha hese

    dephs exis; ha hey can be sysemaically explored; and ha hey may lead o a muli-di-

    mensional reconcepualizaion o genre: such, we hink, are solid ndings o our research.

    Now, more exploraions are on he horizon: he swich rom unsupervised o supervised

    echniques, or insance; or he explici inclusion o semanic daa, which we have so ar

    mosly avoided so as o ocus more sricly on he ormal properies o genres. And hen, a

    he end o i all, he grea challenge o experimenal work: he consrucion o hypoheses

    and models capable o explaining he daa. This sudy is a sep in ha direcion.

  • 7/30/2019 Literary Lab Pamphlet 1

    28/29

    26

    Abou Us

    The Sanord Lierary Lab, direced by Mathew Jockers and Franco Moreti, discusses, de-

    signs, and pursues lierary research o a digial and quaniaive naure. The Lab is open o

    all sudens and aculy a Sanord - and, on a more ad hoc basis, o sudens and aculy

    rom oher insiuions.

    We envisage a variey o projecs, ranging rom disseraion chapers o courses, individ-

    ual or group publicaions, conerence papers and panels, and even shor books. Ideally,

    research will ake he orm o a genuine experimen, and exend over a period o one or

    wo years. On our websie (lilab.sanord.edu) you will nd a lis o our presen aciviies,

    mos o which gaher ogeher several projecs, and are open o urher collaboraion. We

    plan o iniiae wo more experimens in 2010-11, and add anoher wo in 2011-12.

    A he Lab, all research is collaboraive (even hough some oucomes may end up having

    a single auhor). We hold regular group meeings o evaluae he progress o a specic

    experimen, he saus o exising hypoheses, and uure research developmens. (I iner-esed in hese meeings, please conac Jockers or Moreti: as a rule, visiors are welcome.)

    Occasionally, we will have public presenaions o our research, which will be announced

    on our websie under Evens.

  • 7/30/2019 Literary Lab Pamphlet 1

    29/29

    January 2011

    AB