Epicentres

Transcription

Epicentres
1
TobiasBernaisch (JustusLiebigUniversityGiessen)
(paperco-authoredwith
StefanTh.Gries
(UniversityofCalifornia,SantaBarbara;
JustusLiebigUniversityGiessen))
2
Structure
1Theore,calBackground
1.1ThedaFvealternaFon
1.2SouthAsia&SouthAsianEnglishes
1.3LinguisFcepicentres
2Methodology
2.1Corpusdata
2.2Datacoding
3CaseStudies
3.1SourcesofstructuralnaFvisaFoninSouthAsianEnglishes
3.2IdenFfyingtheSouthAsianlinguisFcepicentre
4Discussion
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
1.1Theda,vealterna,on
AlternaFonbetweenthe
double-objectconstrucFon(e.g. JohngaveMaryabook.)
andthepreposiFonaldaFve(e.g. JohngaveabooktoMary.)
Factorsinfluencingthechoiceofonevariantoveranother(cf.e.g.Gries2003;
Bresnan&Hay2008;Schilketal.2013;Bernaischetal.2014):
•  animacyofpaFent/recipient
•  discourseaccessibilityofpaFent/recipient
•  lengthofpaFent/recipient
•  pronominalityofpaFent/recipient
•  semanFcsofpaFent
•  variety
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3
4
1.2SouthAsia&SouthAsianEnglishes
PakistaniEnglish
NepaliEnglish
BangladeshiEnglish
IndianEnglish
SriLankanEnglish
MaldivianEnglish
(takenfromGoogleMaps)
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
5
1.3Linguis,cepicentres
“[A]varietycanberegardedasapotenFal
epicentre if it shows endonormaFve
stabilizaFon (i.e. widespread use, general
acceptance and codificaFon of the local
norms of English) [...] on the one hand,
and the potenFal to serve as a model of
English for (neighbouring?) countries on
theotherhand.”(Hundt2013:185)
(takenfromGoogleMaps)
Epicentreresearchhassofarmainlyrelied
oninferringinterpretaFonsfrom“degrees
ofsimilaritybetweenaspecificdominant
varietyontheonehand(i.e.BriFsh
EnglishorIndianEnglish)andperipheral
varieFesontheother(e.g.SriLankan
EnglishandPakistaniEnglish)”(Hoffmann
etal.2011:261).
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
6
1.3Linguis,cepicentres
Structural similariFes across South Asian
Englishes (SAEs) supporFng the status of
IndianEnglishasanepicentre:
•  Hoffmann et al. (2011) on light-verb
construcFons
•  B e r n a i s c h & L a n g e ( 2 0 1 2 ) o n
presentaFonalitself
Structural differences across SAEs not
supporFng the status of Indian English as
anepicentre:
•  Hundtetal.(2012)onthehypotheFcal
subjuncFve
•  Koch & Bernaisch (2013) on new
ditransiFves
(takenfromGoogleMaps)
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
7
1.3Linguis,cepicentres
“[A]varietycanberegardedasapotenFalepicentreifitshowsendonormaFve
stabilizaFon (i.e. widespread use, general acceptance and codificaFon of the
local norms of English) [...] on the one hand, and the poten,al to serve as a
model of English for (neighbouring?) countries on the other hand.” (Hundt
2013:185)
DaFvealternaFon
Surface-structurechoices
Double-objectconstrucFon
vs.preposiFonaldaFve
NormsconsFtuFng
(variety-specific)models
andguidingsurface-structurechoices
Norms guiding the choice of either
thedouble-objectconstrucFonorthe
preposiFonaldaFve
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
8
1.3Linguis,cepicentres
“[A]varietycanberegardedasapotenFalepicentreifitshowsendonormaFve
stabilizaFon (i.e. widespread use, general acceptance and codificaFon of the
local norms of English) [...] on the one hand, and the poten,al to serve as a
model of English for (neighbouring?) countries on the other hand.” (Hundt
2013:185)
DaFvealternaFon
Surface-structurechoices
NormsconsFtuFng
(variety-specific)models
andguidingsurface-structurechoices
Double-objectconstrucFon
H e g a v e h i s
Hegaveher
d a u g h t e r t h e
abook.
freedomtocome
homelate.
Factors:
Factors:
recipient=pronominal
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
recipient
recipient
paFent
paFent
≤5words
=non-pronominal
>3words
=abstract
9
1.3Linguis,cepicentres
“[A]varietycanberegardedasapotenFalepicentreifitshowsendonormaFve
stabilizaFon (i.e. widespread use, general acceptance and codificaFon of the
local norms of English) [...] on the one hand, and the poten,al to serve as a
model of English for (neighbouring?) countries on the other hand.” (Hundt
2013:185)
Surface-structurechoices
EpicentreidenFficaFon
NormsconsFtuFng
(variety-specific)models
andguidingsurface-structurechoices
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
10
2.1Corpusdata
SouthAsianVarieFesofEnglish(SAVE)CorpusfeaturingsixnaFonalcomponents
with3mwordsofnewspaperlanguagepervariety(cf.Bernaischetal.2011)
NewssecFonfromtheBriFshNaFonalCorpusforBriFshEnglish(BrE)referencedata
Variety
Sources
URLs
BangladeshiEnglish
DailyStar
NewAge
hlp://www.thedailystar.net
hlp://www.newagebd.com
IndianEnglish
TheStatesman
TheTimesofIndia
hlp://www.thestatesman.net
hlp://Fmesofindia.indiaFmes.com
MaldivianEnglish
DhivehiObserver
MinivanNews
hlp://www.dhivehiobserver.com
hlp://www.minivannews.com
NepaliEnglish
NepaliTimes
TheHimalayanTimes
hlp://www.nepaliFmes.com
hlp://www.thehimalayanFmes.com
PakistaniEnglish
DailyTimes
Dawn
hlp://www.dailyFmes.com.pk
hlp://www.dawn.com
SriLankanEnglish
DailyMirror
DailyNews
hlp://www.dailymirror.lk
hlp://www.dailynews.lk
Bri,shEnglish
NewssecFonoftheBriFsh
NaFonalCorpus(BNC)
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
11
2.2Datacoding
1381exampleswithGIVEwereannotatedaccordingtothefollowingvariables
Variable
Descrip,on
Variants
VARIETY
thevarietyfromwhichthe
examplewastaken
PAPER
thenewspaperfromwhich
theexamplewastaken
LOGLENDIFF
loggeddifferencebetween
lengthofrecipientand
lengthofpaFent
[numericvalue]
RECANIMACY&PATANIMACY
animacyofrecipientand
paFent
animatevs.inanimate
RECACCESSIBILITY&PATACCESSIBILITY
discourseaccessibilityof
recipientandpaFent
givenvs.new
RECPRONOMINALITY&
PATPRONOMINALITY
realisaFonofrecipientand
paFentinpronominalor
non-pronominalform
pronounvs.np
PATSEMANTICS
semanFcclassofpaFent
abstractvs.concretevs.informa<onal
TRANSITIVITY
verb-complementaFonal
palernofGIVE
ditransi<vevs.preposi<onalda<ve
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.Methodologicalpreliminaries:theMuPDARapproach
12
•  newregression-basedapproachfromthedomainoflearnercorpusresearch
called MuPDAR (MulFfactorial PredicFon and DeviaFon Analysis using
Regression,seeGries&Deshors2014):
1.  generateaconcordanceofthephenomenonofinterestfromNSandNNS
data and annotate for predictors P1-n known/hypothesized to affect the
phenomenon;
2.  fitaregressionR1tomodelthephenomenonasafuncFonofP1-nintheNS
dataonlyandcheckR1’sclassificaFonaccuracy;
3.  ifR1’sclassificaFonaccuracyisgood,applyR1totheNNSdatato,foreach
case,getapredicFonof‘whataNSwouldhavedonehere’;
4.  compare whether the NNS made the predicted NS choices and fit a
regression R2 to model where and how much the NNS made nonnaFvelike/non-idiomaFcchoices.
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.1Casestudy1:Iden,fyingfactorsofstructuralna,visa,oninSAEs
13
•  extensionofMuPDARtothecomparisonofBrE(asthe‘target’variety,NSin
the above) to indigenised varieFes (as the ‘learner’ varieFes, NNS above):
what factors are responsible for indigenised-variety speakers making nonBrEchoices?
•  with the annotated predictors on the BrE data, R1 is created and its
classificaFon accuracy is evaluated; crucially, R1 is a mixed-effects model
taking the relatedness of data points from the two BNC parts into
consideraFon;
•  R1 is applied to the indigenised-variety speakers to predict BrE speakers’
choices and a variable called VARIETYSPECIFICITY staFng whether non-BrE
speakersmadeBrEchoicesornotiscreated;
•  withtheannotatedpredictorsontheindigenised-varietydata,R2iscreated
for VARIETYSPECIFICITY; crucially, R2 is a mixed-effects model taking the
hierarchical structure of the corpus data into consideraFon (VARIETY/
NEWSPAPER,seeGries2015);
•  thedegreeofhowmuchthenon-BrEspeakers’choicesdifferedfromthose
theBrEspeakersmadewasalsocreated.
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.1Casestudy1:Iden,fyingfactorsofstructuralna,visa,oninSAEs
14
Results
•  R1resultedinaverygoodclassificaFonaccuracyof93.7%ontheBrEdata
(***belerthanchance)
•  R1resultedinagoodpredicFonaccuracyof77%onthenon-BrEdata
(***belerthanchance)
•  VARIETYSPECIFICITYvalueswerecomputed:
•  iftheSAEspeakermadetheBrEchoice,
VARIETYSPECIFICITY=0
•  iftheSAEspeakerdidnotmaketheBrEchoice,
VARIETYSPECIFICITY=0.5–predictedprobabilityofpreposiFonaldaFve
•  thus,ifVARSPEC>0,SAEuserusedaprep.daFve,butaBrE
speakerwouldhavechosenaditransiFve
•  thus,ifVARSPEC<0,SAEuserusedaditransiFve,butaBrEspeaker
wouldhavechosenaprep.daFve
•  R2resultedinagoodclassificaFonaccuracyof77.2%
(***belerthanchance)
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.1Casestudy1:Iden,fyingfactorsofstructuralna,visa,oninSAEs
•  SAE speakers typically make
BrE-like choices: most points
arearoundy=0;
•  t h e m o r e t h e l e n g t h
differencegivesaclue(i.e,the
furtherxisfrom0),themore
BrE-liketheirchoicesare;
•  whenthepaFentisgiven,SAE
s p e a k e r s m a k e n o n - B r E
choicesequallymuch(see×);
•  whenthepaFentisnew,they
are much more likely to
choose non-BrE-like preposiFonaldaFves(see×);
•  thus, compared to BrE, the
strength of the cue ‘new
paFent’ is stronger for prep.
daFvesinSAEs.
15
ditr.:Shegave
himREC/givenabookPAT/new.
prep.dat.:Shegave theFcketPAT/givento amanREC/new.
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.1Casestudy1:Iden,fyingfactorsofstructuralna,visa,oninSAEs
•  SAE speakers typically make
BrE-like choices: most points
arearoundy=0;
•  t h e m o r e t h e l e n g t h
differencegivesaclue(i.e,the
furtherxisfrom0),themore
BrE-liketheirchoicesare;
•  when the recipient is given,
SAE speakers make non-BrE
choicesequallymuch(see×);
•  when the recipient is new,
they are much more likely to
choose non-BrE-like preposiFonal daFves (see ×) – it
seems in fact as if the cue
‘newrecipient’forprep.dat.is
strongerforSAEsthanforBrE.
16
ditr.:Shegave
himREC/givenabookPAT/new.
prep.dat.:Shegave theFcketPAT/givento amanREC/new.
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.1Casestudy1:Iden,fyingfactorsofstructuralna,visa,oninSAEs
17
ditr.:Shegave
himREC/givenabookPAT/new.
prep.dat.:Shegave theFcketPAT/givento amanREC/new.
•  SAE speakers typically make
BrE-like choices: most points
arearoundy=0;
•  when the recipient is a
p ro n o u n , th en th e S AE
choices are BrE-like (see •),
esp.whenthepaFentisnew;
•  when the recipient is lexical,
SAE speakers are more likely
than BrE speakers to use
preposiFonaldaFves(see•).
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.1Casestudy1:Iden,fyingfactorsofstructuralna,visa,oninSAEs
18
•  Note:someSAVEcomponents(Ind,Pak,SL)aremuchmorehomogeneousthan
others(Ban,Mal,Nep)
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.2Casestudy2:ExploringepicentersinSAEs
19
Following Hundt (2013: 185), epicentres have “the potenFal to serve as a
modelofEnglishfor(neighbouring?)countries”.
Proposed operaFonalisaFon: model = set of norms governing structural
choices; thus, the similarity of the norms of varieFes to those of an assumed
epicentrewillreflecthowlikelytheassumedepicentreisanepicentre.
Theproposedmethodisabolom-upextensionofMuPDAR:
•  eachofthevarieFesstudiedherewillbeassumedtobetheepicentre,i.e.
yieldR1-basedpredicFonsforallothervarieFes:
•  coarse-grainedapproach:%sofstructuralchoicesaspredictedbythe
assumedepicentre;
•  fine-grainedapproach:absolutedeviaFonsfromallvarieFes’users’
choicesfromtheassumedepicentre;
•  then,eachofthesestaFsFcswassummedupforeachassumedepicentre
andthesumswereploled.
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
3.2Casestudy2:ExploringepicentersinSAEs
•  BrEbehavesquitedifferently
fromthe(clusterof)SAEvarieFes
•  inthecoarse-graineddata
•  inthefine-graineddata
•  incombinaFon,theresultspoint
toIndEastheepicentre:
•  inthecoarse-grainedplot,
IndEisnarrowlybestedby
PakE,but...
•  inthefine-grainedplot,IndE
isthevarietyfromwhichthe
othersarepredictedbest(w/
thesmallestsumofdevia-
Fons).
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
20
4Discussion–Epicentres
IndianEnglishasthelinguisFcepicentreforSouthAsianEnglishes?
CriteriaforalinguisFcepicentre:
yes
•  endonormaFvestabilisaFon?
•  modelforothervarieFesinitsvicinity? yes
Desiderata:
•  comparablestudiesof
otherphenomena/alternaFons
•  diachronicdata
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
21
22
4Discussion–Methods
•  ‘what would a speaker of a/the historical input variety have done’ in the
structuralsituaFoninwhichtheESLspeakerfindsherselforhimself?
•  MuPDAR can be used exploraFvely to examine potenFal epicentral
configuraFons
•  the hierarchical structure of the corpus data is taken into account by using
mulF-levelmodelingintheregressionmodeling
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
23
References
Bernaisch, Tobias and Claudia Lange. 2012. “The typology of focus marking in South Asian Englishes”. Indian
Linguis<cs73(1–4):1–18.
Bernaisch, Tobias, Christopher Koch, Joybrato Mukherjee and Marco Schilk. 2011. Manual for the South Asian
VarieFes of English (SAVE) Corpus: Compila<on, Cleanup Process, and Details on the Individual Components.
Giessen:JustusLiebigUniversity.
Bernaisch, Tobias, Stefan Th. Gries and Joybrato Mukherjee. 2014. “The daFve alternaFon in South Asian
English(es):modellingpredictorsandpredicFngprototypes”.EnglishWorld-Wide35(1):7–31.
Bresnan, Joan and Jennifer Hay. 2008. “Gradient grammar: an effect of animacy on the syntax of give in New
ZealandandAmericanEnglish”.Lingua118:245–259.
Gries,StefanTh.2003.“Towardsacorpus-basedidenFficaFonofprototypicalinstancesofconstrucFons”.Annual
ReviewofCogni<veLinguis<cs1:1–27.
Gries,StefanTh.2015.ThemostunderusedstaFsFcalmethodincorpuslinguisFcs:MulF-level(andmixed-effects)
models.Corpora10(1).95-125.
Gries,StefanTh.andSandraC.Deshors.2014.UsingregressionstoexploredeviaFonsbetweencorpusdataanda
standard/target:twosuggesFons.Corpora9(1).109–136.
Hoffmann,SebasFan,MarianneHundtandJoybratoMukherjee.2011.“IndianEnglish–anemergingepicentre?A
pilotstudyonlightverbsinweb-derivedcorporaofSouthAsianEnglishes”.Anglia129(3–4):258–280.
Hundt,Marianne.2013.“ThediversificaFonofEnglish:old,newandemergingepicentres”.InDanielSchreierand
MarianneHundt,eds.EnglishasaContactLanguage.Cambridge:CambridgeUniversityPress,182–203.
Hundt,Marianne,SebasFanHoffmannandJoybratoMukherjee.2012.“ThehypotheFcalsubjuncFveinSouthAsian
Englishes:localdevelopmentsintheuseofaglobalconstrucFon”.EnglishWorld-Wide33(2):147–164.
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes
24
References
Koch, Christopher and Tobias Bernaisch. 2013. “Verb complementaFon in South Asian English(es): the range and
frequencyof‘new’ditransiFves”.InGisleAndersenandKrisFnBech,eds.EnglishCorpusLinguis<cs:Varia<on
inTime,SpaceandGenre–SelectedPapersfromICAME32.Amsterdam:Rodopi,69–89
Schilk, Marco, Joybrato Mukherjee, Christopher F.H. Nam and Sach Mukherjee. 2013. “ComplementaFon of
ditransiFveverbsinSouthAsianEnglishes:amulFfactorialanalysis”.CorpusLinguis<csandLinguis<cTheory
9(2):187–225.
T.Bernaisch:EpicentralConfiguraFonsinSouthAsianEnglishes