40 1

Deardelegates,Welcometothe2017UniversityofEdinburghLinguisticsandEnglishLanguagePostgraduateConference!Runningcontinuouslysince1994,theLELPGCisaplaceforpostgraduateresearcherswithinEdinburghandbeyondtoshowcasetheirresearchandgetusefulfeedback.Wehopethatyouenjoythediverseprogrammeofworkthatwehaveputtogetherfromthemanyhighqualityabstractsthatwereceivedthisyear.Ifyouhaveanyquestionsduringtheconference,pleasedonothesitatetocontactamemberoftheCommittee–wewillbepresentaroundtheconferenceeachdaytokeepthingsrunningsmoothly.Enjoy!Bestwishes,TheconferencecommitteeMirjamEiswirth(Chair)•WenjiaCai•JonCarr•AdamScottClarkRuthCorps•TobyDathan•VictoriaDickson•AndresKarjusTianLi•MaddieLong•ChristineLugrine•ErnisaMarzuki

RachelMoyer•EvaSchnelten•MirandaStocks•BetülSedaUstaTomWood•YifeiYang

------

CONTENTS

Programme 2Abstracts(bysession) 5Usefulinformation 40

2

PROGRAMMEDay1:Monday29thMay

MSc

WORK

SHOPS 09.30-10.15 Panelsession:thedissertation

experienceRobTruswell&CandiceMathers

10.15-11.00 PresentingatConferences IainDavidson11.00-11.45 Researchmethodologies HannahRohde

11.45-12.00 BREAK

PLENARY 12.00-13.00 ‘Languagecontainssociety’:Hong

Kong’sumbrellamovementthroughBenveniste’sLastLectures(andvice-versa)(p.6)

Prof.JohnJoseph

13.00-14.00 BREAK

SESSION1

14.00-14.25 Acloserlookatsyntaxofclausal

complementsoffactiveverbsinJordanianArabic(p.7)

MarwanJarrah

14.25-14.50 CallingFinnishspeakers:Evidenceforspeakerandhearersyntaxfromdiscourseparticles(p.8)

AnnaHolingsworth

14.50-15.15 ThesyntaxofsententialnegationinAbhaArabicdialect(p.9) AliAlzayid

15.15-15.40 Doubletopicalisationsyntaxdoesconstrain!(p.9) SherryYongChen

15.40-15.50 BREAK

SESSION2

15.50-16.15 Practicedlanguageattitudes:Acase

ofmothertongueeducationinTaiwan(p.11)

Chia-YingYang

16.15-16.40 Emergingself-identitiesandemotions:AqualitativecasestudyoftenSaudistudentsintheUK(p.12)

OunAlmesaar

16.40-17.05 Metadiscourseintheclassroom:Acomparativeanalysisofnativeandnon-nativeEFLteachers(p.12)

XinxinWu

17.05-17.15 BREAK

SESSION3

17.15-17.40 Across-linguisticapproachtoadpositionsinthirdlanguageacquisition(p.13)

SakineÇabuk

17.40-18.05 Thinkaloudprotocols:AwindowintothemindsofhowEnglishasaforeignlanguagestudentsprocessfeedback(p.13)

SharifaAlHarrasi

NOTES

47

3

18.05-18.30 TheinfluenceoforalreadingonKoreanlanguageproficiency:Acasestudy(p.14)

JihyeEo

18.30 CONFERENCEDINNER Day2:Tuesday30thMay

SESSION4

09.00-09.25 Scarystatisticsandhistoricallinguistics(p.15)

DaisySmith

09.25-9.50 Language,identityandcommunity:Theuseofthird-personneuterpossessivesofearlymodernpreachers(p.16)

HiroshiYadomi

9.50-10.15 DiscourseandtopicalizationinoldEnglishsubordination(p.17) SergioLopezMartınez

10.15-10.25 BREAK

SESSION5

10.25-10.50 Mechanismsofcognitivecontrolinbilinguals:Theempiricalgeneralizabilityofthe‘bilingualadvantage’(p.17)

MichelaBonfieni

10.50-11.15 EffectsofconceptualaccessibilityandsimilarityinsimplesentenceproductioninMandarin(p.18)

YangziZhou,HollyBranigan,&MartinPickering

11.15-11.40 SyntacticprimingandanimacyeffectsinRussian-speakingchildren(p.19) AlinaKonradt

11.40-12.05 Parsingwh-questions:EvidencefromL1-Greekadultsandimplicationsforbilinguallearning(p.20)

KaterinaPantoula

12.05-13.05 LUNCH&POSTERSESSION

Theinfluenceoftemporalcontextonthe

productionoftemporalmorphologyinL2speakersofEnglish(p.21)

QingyuanGardner,VickyChondrogianni,&HollyBranigan

Indirectnessinthefaceofculturalcompetency(p.23) MayOuma

Phoneticpreaspirationofword-finalvoicelessfricativesinNorthEastScotland(p.23)

DavidA.J.Warren

Exploringyounglearners’foreignlanguageanxietyinChina(p.24) XiaoyiHu

SESSION6

13.05-13.30 Iteratedlearningoptimizesforsimplicity

(p.25) JonCarr

13.30-13.55 Populationdynamicseffectsontheevolutionofcommunicativeconventions(p.26)

JoséSegoviaMartín&MónicaTamariz

13.55-14.20 Squaringaline:Howsimilararegradableadjectivestoquantifiers?(p.27) FaustoCarcassi

creating graph-based annotations, where all meaningful parts of speech aredividedintocategories,forinstance,“state”,“religion”,“values”,etc.Theanalysisis made in the context of the “codes” of the given countries (in the light ofhistorical,cultural,andpoliticalexperienceofthesociety).PLENARY2 WEDNESDAY31STMAY,15.00-16.0015.00-16.00The acquisition of case-marking and word order in heritage speakers:GreekheritagechildreninNewYorkCityVickyChondrogianni•UniversityofEdinburgh

Heritage speakers are bilingual speakers who acquire the minorityheritage language frombirthathomeand themajority languageof thesocietythrough schooling and the wider community (e.g. Montrul, 2016; Polinsky &Kagan, 2007). Heritage speakers from immigrant communities, in particular,typicallyhaveweakerabilitiesintheirheritagelanguageandhavebeenshownto perform differently from monolingual speakers of the same language whogrowupinthecountryoforigin.Areasof languagethatareusuallyvulnerableare those where crosslinguistic differences between the minority and themajority languages occur.Much research on heritage speakers has focused onadults with fewer studies examining the factors that affect heritage bilingualchildren’sdevelopmentoftheirheritagelanguage.

In this talk, I present a series of production, online and offlinecomprehension experiments examining the linguistic (case ambiguity, wordorder) and extralinguistic (heritage language proficiency and use) factors thataffectheritagechildren’sabilitytointegratewordorderandcase-markingcuesin sentence production and comprehension when the dominant language(English)offersevidenceforwordordercuesbuttheheritagelanguage(Greek)utilises case-marking cues or conflicting word order cues. The results arediscussedwithincurrenttheoriesofheritagelanguageacquisitionandbilingualdevelopment.

46

4

14.20-14.45 Iseewhatyoudidthere:Theroleoficonicityintheacquisitionofsigns(p.28) AshaStewart

14.45-14.55

BREAK

SESSION7

14.55-15.20 Freeadjunctsanddiscoursestructure(p.

29) JamesE.M.Reid

15.20-15.45 Constructiongrammarandlexicalpolysemy:Acasestudyoftheverbfeedintwoargumentstructureconstructions(p.30)

IvanaKralikova

15.45-15.55 BREAK

SESSION8

15.55-16.20 Ascientificrealiststancetowards

sociolinguisticmetatheory(p.31) JohannesWoschitz16.20-16.45 TheUnitedStatesof‘Merica:Indexicality

andTwitterhashtaguse(p.32) RachelMoyer16.45-17.10 ThechangingSwedish/iː/vowel:Evidence

fromthreecentralSwedishcities(p.33) FabienneWesterberg

17.30-19.30 POTLUCKPICNICINTHEMEADOWS 20.00-21.30 LANGUAGETOUROFEDINBURGH Day3:Wednesday31stMay

SESSION9

09.00-09.25 Linguisticnicheoracademicniche?Theroleoflanguage-externalbiasesinlanguageevolution(p.34)

JonasNölle

09:25-09.50 Theculturalco-evolutionoflanguageandmindreading(p.35)

MariekeWoensdregt,SimonKirby,ChrisCummins,&KennySmith

09.50-10.15 InvestigatingeffectsofmorphologicaltypesandtheirwordformationpatternsinSLA(p.36)

SvenjaWagner

10.15-10.25 BREAK

SESSION10

10.25-10.50 Thedevelopmentoftensemorphology

byWelsh-EnglishbilingualchildrenwithandwithoutSpecificLanguageImpairment(SLI)(p.37)

HyowonKwon&VickyChondrogianni

10.50-11.15 Instructedvs.uninstructedbilinguals:Theroleplayedbymetalinguisticawarenessinthirdlanguageacquisition(p.38)

FrancescaD’Angelo

Ethnographyhasitsroot inanthology(Fiona&Creese,2015). Inthisresearch,linguisticsethnographywillbeadopted,where Iobservedandparticipate inamoduleinaMaster’sprogrammeataUKuniversityfromintheSpringSemesteroftheacademicyear2016/17.FouroftheChinesestudentsinthismodulewerethe key participants, and what they said in class was audio recorded as theprimarysourceofdatainthisresearch.Andfollow-upinterviewswillbecarriedoutintheendwitheachkeyparticipanttofigureoutthereasonswhytheyuseChinese/English in class. The data collection for classroom conversation hasalreadyfinished,anditispossiblethatpartoftheinterviewscanbecarriedoutbeforetheendofMay;therefore,Iwillbeabletopresentsomeofthedatainthepresentation. 14.20-14.45Strategies of persuasion and argumentation in political rhetoric onexample of the inaugural addresses of American, British, Byelorussian,German,andRussianElenaAfromeeva•Friedrich-SchillerUniversityJena&BournemouthUniversity

The research pertains to Gerry Philipsen`s ethnographic study ofcultural communication (1981), especially on its` proposition that one of themajorfunctionofcommunicationistoconstructasenseofsharedidentity.Thus,it is postulated that every ethnic group possesses a distinct “code” which isdefinedas“historicallytransmitted,sociallyconstructedsystemofsymbolsandmeanings, primes and rules, pertaining to communicative conduct” (Philipsen1992:124,Philipsen2010).Thestudy investigates thedevelopmentofpoliticalrhetoric in the abovementioned countries in the light of their "code", i.e. inconnetion with their unique historical background and cultural peculiaritieswhichcreateanexclusivepolitical context.For instance, the researchassumesthat even identical rhetorical means will attain a wide range of connotationsdepending on cultural, historical andpolitical development of the society. Thestudyprovesthedependenceoflinguisticandstylisticdevices,whichembellishpresidentialoratory,onculturalandhistoricalcomponentsofthe“codes”whichtheypertainto.Theresearchinvestigatestheconnectionoflanguageandlinguo-stylistic devices which politicians use to manipulate public opinion andconstruct a sense of shared identity with cultural, historical, and politicalbackgroundtheyrelyon,inordertoapproachanidealimageofapolitician.Thestudy analyses variations of rhetorical patterns across cultures and theirreliance on a unique "code" of a specific community. The central idea of thestudy is rooted in the proposition that in order to be more persuasive andeloquent,presidentsappeal to the featuresof their “codes”andapply identicallinguisticandstylisticmeansindifferentconnotations.

Central are the notions of persuasion and argumentation in politicalcommunication with special focus on the inaugural addresses delivered byAmerican,British,Byelorussian,German,andRussianpoliticalleaders.

The objectives of the study can be practically demonstrated in thecorpus of the political speeches which is aimed to be created for the publicusage. The practical analysis ismade bymeans of GraphAnno programme for

45

5

11.15-11.25 BREAK SESSION11

11.25-11.50 Onthestructureofpossessiveħaggin

freestatenominalsinHailiArabic(p.39)

EisaAlrasheedi

11.50-12.15 ThenounphraseincentralKurdish:AprojectionofD(DP)notNum(NumP)(p.40)

RebwarTahir

12.15-12.40 Variationalmodelsoflanguageacquisition:Thebestisstillnotgoodenough(p.42)

CallumHackett

12.40-13.30 LUNCH

SESSION12

13.30-13.55 Acorpus-lexicographicaldiscourseanalysisoftheverb‘destroy’(p.43)

EmmaFranklin

13.55-14.20Master’sstudents’languageuseoftheirfirstandsecondlanguagesinclass:Asmall-scalecasestudy(p.44)

MingNi

14.20-14.45

StrategiesofpersuasionandargumentationinpoliticalrhetoriconexampleoftheinauguraladdressesofAmerican,British,Byelorussian,German,andRussian(p.45)

ElenaAfromeeva

14.45-15.00 BREAK PLENARY 15.00-16.00 Theacquisitionofcase-markingand

wordorderinheritagespeakers:GreekheritagechildreninNewYorkCity(p.46)

DrVickyChondrogianni

16.00-17.30 WINERECEPTION 18.30-19.30 PUBSANDWUGSPUBEVENING

-----

AlltalkswilltakeplaceinLG.09inDavidHumeTower(seemaponpage47).

BREAKSwillincluderefreshmentsoutsideLG.09.LUNCHwillbeinthe

commonroomonthe7thfloorofDugaldStewartBuilding.

Thereisliftaccesstoallfloors.

‘killees’incontemporaryBritishEnglish.Informedbytheobservationthattheverb‘destroy’isuseddifferently

for humans and animals, namely as a ‘killing’ termwhen applied to dogs andhorses,forexample,butnothumans,aCPAanalysiswascarriedoutonrandomsamples of concordance lines of ‘destroy’ from the BNC and the PPPP corpus.Thepreliminaryresultsindicatethattherearefoursensesoftheverb‘destroy’,asmarkedbytheirpatterns:twopertainingtodamageasinflicteduponphysicalobjects and abstract entities, respectively; one for referring to the defeat of ahuman group (e.g. in battle, sport, etc.); and one for denoting the killing ofunwanted or dangerous animals. The patternswere found to be stable acrossthetwocorpora,butthePPPPcorpussamplefeaturedmoreexamplesofanimal-relateddestruction (e.g. of animal homes rather thanhumanones) and raisedquestionsaboutthedistinctionbetweenhumansand‘nature’.Moreimportantly,it was found that by delimiting verb patterns using CPA, it is possible toempirically show when, and in what contexts, an entity is conceptualised asbelonging to another category. In this case, it was found that some animals –particularly wild animals – are subtly objectified by their roles in ‘destroy’constructions,asdemonstratedusingprecisedemarcationofverbpatternsandtheirarguments.

13.55-14.20Master’sstudents’languageuseoftheirfirstandsecondlanguagesinclass:Asmall-scalecasestudyMingNi•UniversityofSterling

Withthedevelopmentof internalizationandglobalization, there isanincreasingnumberofstudentschoosingtofurthertheirstudyabroad(Ilievaetal.,2004).Forinstance,theUKhoststhesecondlargestnumberofinternationalstudents in in the world (UK Council for International Student Affairs, 2015),with 438,010 international students during the academic year of 2015/16,accountingfor19.2%ofthetotalnumberofstudentsstudyingintheUK(HigherEducation Statistics Agency, 2017). And within these overseas students, thelargest proportion of them comes from China (Higher Education StatisticsAgency, 2016). For these Chinese students, English is the classroom languagebutnottheirfirstlanguage.Sohowistheiruseoflanguagesinclass?

In the presentation, I will report on an ongoing pilot for my PhDproject,whichlooksatChineseMaster’sstudents’languageuseoftheirfirstandsecond languages (i.e. Chinese and English) in class in the UK. Theseinternationalstudents’languageuseinsideclassroomhasdrawntomyattentionbecause I have noticed some interesting phenomena on some non-nativestudents’useoftheirfirstandsecondlanguagesinandafterclass(oneexamplecouldbetheytendtospeakChinesewithanotherChinese,evenifthereisanon-Chinesearound),andwouldliketofurtherinvestigateit.Theresearchquestionsofthisstudyinclude:

1. Underwhat circumstances do Chinese students use each language in

class?2. HowdoChinesestudentsexplaintheirchoiceoflanguage?3. Howtheprocessandmethodsofdatacollectionandmyparticipation

canbeimprovedformyPhDproject?

44

6

PLENARY1 MONDAY29THMAY,12.00-13.0012.00-13.00‘Languagecontainssociety’:Hong Kong’s Umbrella Movement interpreted through Benveniste’s LastLectures(andvice-versa)ProfJohnE.Joseph•UniversityofEdinburgh

Intheautumnof2014aprotestmovementwaslaunchedinHongKongagainst the government’s plan to constrain the electoral process for the ChiefExecutive.Througha complex chainof linguistic and semiotic associations theprotests quickly came to be called the Umbrella Movement, in Hong KongCantonese��ze1daa2“Chater”,literally“coverhit/beat”:ChaterRoadhavingbeenthelocationofearlyprotests,anditsfirstcharacter,�ze1,beingalsotheword for umbrella. Protesters had used their umbrellas to cover themselvesfrom the tear gas and baton attacks launched at them by police, hence thesignificance of the literalmeaning of ze1 daa2,which is opaque to speakers ofPutonghua.EvenmainlandCantonesecannotfullydecodethesignificanceofthemovement’sname.

Émile Benveniste (1902-1976) gave his last series of lectures in theCollègedeFrance in1968-1969.TheseDernières leçonsremainedunpublisheduntil 2012, and future historians might be forgiven for imagining that theUmbrellaMovementof2014wasdirectlyinspiredbythem.Itwasnot,ofcourse,andyetthemovement’ssemiologyprovidesevidenceforBenveniste’skeyideas.Benveniste laid out a semiological theory extending Peirce’s idea of the‘interpretant’ –whichmaintains that any sign ismeaningful only through theintermediaryofanothersign– to includethe languagesystemasconceivedbySaussure.ForBenveniste,othersignsystemsdependonlanguagetoactastheirinterpretant,andthesesystemsincludesocietyitself.

Theinterpretingsystemfurnishesthebaseoftherelationsthatpermit

whatisinterpretedtodevelopintoasystem.Thebaseisfurnishedbythelanguage: for example, the pronoun system, I/you versus he/she/it.Without this linguistic distinction that introduces the relationship ofdialogueandthatofotherness,nosocietyispossible.

Betweenthetwosystems,linguisticandsocial,thereisnostructuralcorrelation. The relation can only be semiological, namely a relation ofinterpretanttointerpreted,excludinganygeneticrelation.

Languagecontainssociety.Thelanguagecanbestudiedonitsown,asaformalsystem,without

taking account of society. The inverse is not the case. Society and therepresentations that govern it cannot be described outside linguisticrealisations.

Only what language denotes is social. (Benveniste 2012: 79, mytranslation)

How the umbrella took on the symbolism it did in Hong Kong in 2014 is asnebulous as any social creation of meaning. But deeply involved in it, asinterpretants, were the languages Putonghua and Cantonese – on the level oftheirwordsforumbrella(Putonghua�săn,HongKongCantonese�ze1),butalso on that of the whole existence of Cantonese and Putonghua as languagesystems, rather than as dialects of oneChinese language, or Putonghua as the

betweenachild’s input languageand theirproductionofRoot Infinitivesgivessuchmodelsmoreexplanatorypotential thanotherprominentaccountsof thephenomenon(e.g.,Rizzi,1994andWexler,1998).Iwillshowthatthesemodelscanberefinedevenfurtherbyusingmorenuancedsetsofparameters,sothatthesyntacticprocessesunderlyingRoot Infinitives,aswellas importantcross-linguisticdifferences,canbeevenmorerobustlycaptured.

Nonetheless,acoredifficultyfacedbyVariationalModels–andbythePrinciples&Parametersapproachtogenerative linguisticsasawhole– is thatputativeparametersthatarederivedfromcomparativedata(like ‘nullsubject’or ‘head position’) tend to encode descriptive facts in a technical system thatlooksexplanatorywithoutactuallyaddinganytheoreticalinsight.Thesourceofthisdifficultyissimplythatthereisaninfinitevarietyofbinaryparametersthat canbepositedtoaccountforcross-linguisticvariationandnounderlyingtheorytoconstrain them. Ina similarvein toBoeckx (2011), Iwill suggest that therecannot be any such constraining theory and that even the most attractiveparameter-based models of acquisition and adult grammars must necessarilyfallshort. SESSION12 WEDNESDAY31STMAY,13.30-14.45 13.30-13.55Acorpus-lexicographicaldiscourseanalysisoftheverb‘destroy’EmmaFranklin•LancasterUniversity

Thispaperpresentsaworkinganalysisoftheverb‘destroy’asusedinmodern British English, taking Hanks’ (2004, 2013) Corpus Pattern Analysis(CPA)approachasamethodologicalbasis.Todate,CPAhasonlybeenappliedincorpus lexicography and similarNatural LanguageProcessing tasks, but it hasthepotentialtoextendcurrentdiscourseanalysismethods.Assuch,thispaperdemonstrates an early attempt at a CPA-assisted, corpus-lexicographicaldiscourse analysis, using evidence from both a general corpus of English (theBritish National Corpus) and a specific, animal-themed corpus constructed byresearchers on the ‘People’, ‘Products’, ‘Pests’ and ‘Pets’ (PPPP) project (SealeyandPak, forthcoming).The study ispartof a largerproject–myPhD–whichseekstoestablishwhataCPAanalysiscanrevealaboutattitudestoanimalsbyexaminingtheirpositionswithin‘killing’verbconstructions.

CPA is underpinned by Hanks’ (2013) Theory of Norms andExploitations (TNE), a practical and empirical theory of language in the Neo-Firthian tradition, which posits that meaning is context-dependent and isinextricablylinkedtoform(cf.Sinclair(1991)).CPAinvolvesmanuallyanalysingthe co-texts of words (especially verbs), and from these concordancesidentifying syntagmatic patterns associated with senses of meaning. ThisprocessissimilartotheapproachtakenintheCOBUILDproject(Francisetal.,1996;Sinclair,1987),andsubsequentlyPatternGrammar(HunstonandFrancis,2000), but it places a greater emphasis on the semantic values of argumentsthanpreviouscorpus-drivenapproachestolanguagepatterns.Thismakesitanidealmethod for empiricallydemonstrating the typesof entities–or semantictypes – portrayed as participants in particular events. In the case of my PhD,these are the humans, animals and other entities that constitute ‘killers’ and

43

38

7

languageandCantoneseasadialect.Thestatusofthelanguagesreproduces,andallows speakers to perform, the political status of Hong Kong vis-à-vis thePeople’s Republic of China. Also consideredwill be Benveniste’s views on therole ofwriting in the ‘auto-semiotisation’ of language, and its ability to ‘short-circuit’ spoken language, aswell ashis insistence that society is, not reflected,butcontained,inlanguage.SESSION1 MONDAY29THMAY,14.00-15.4014.00-14.25AcloserlookatsyntaxofclausalcomplementsoffactiveverbsinJordanianArabicMarwanJarrah•NewcastleUniversity

Therearetwopossibleaccountsofimpossibilityof(subject)extractionout of clausal complements of factive verbs. On the one hand, factivecomplementshavebeenassumedtobeCPsheadedbyanullD°,somethingthatturnssuchclauses intostrong islands(cf.Kiparsky&Kiparsky1970,Manahlot1977,andOuhalla2005).Ontheother,theleftperipheryoffactivecomplementsobtains an operatorwhich blocksmovement out of these clauses (cf.Melvold1991, Hegarty 1992, Zubizaretta 2001, and Haegeman 2012). In this paper, Ipresent these twoaccounts andargue that theDPapproach is superior to theoperatorapproachinJordanianArabic(JA).

Firstly,accordingtotheoperatorapproach,A-barmovementtotheleftperipheryof factivecomplements isnotpermitted(Grimshaw1990,Makietal1999, and Zubizaretta 2001) or severely restricted; only Discourse-linkedelementscanbeextracted(Haegeman2012).Assuch,focalization,amongotherthings,wouldnotbepermittedwithinJAfactivecomplements,contrarytofact.Considerthefollowingsentence:

(1)ʔabuu-Iħizin/ʔistaʁrab/nasa/ʕirif ʔinn-haBINIT

father-myregretted/gotsurprisedof/forgot/knew that-3SF girlʔaχoo-isaragbrother-mystole.3SM

'Myfatherregretted/gotsurprisedof/forgot/knewthatitthatitwasagirlthatmybrotherhitatthemarket.’

Grammaticality of (1) is ample evidence that fronting to the leftperipheryoffactivecomplementsisnotdisallowedinJA.(1)ishardtoaccountforassumingthatthereisanoperator(intheleftperipheryoffactiveclausesinJA) blocking A’-movement instances to the left periphery, as all A-barmovementsarepermitted.Now,considerthefollowingill-formedexamplethatshows that no extraction (of arguments or adjuncts) is allowed to thematrixclause:

(2)*miinz-zalamih/weenʔabuu-I ħizin/ʔistaʁrab/nasa/ʕirifwhichman/whenfather-myregretted/gotsurprisedof/forgot/knewʔinn-hasarag-t ʔis-siyaarah

demonstratives(4),isthesyntacticrealizationofdefiniteness,servingasimilarfunctiontothemaindefinitemarker–ekein(2).

(4) ew esp-an-ethathorse-PL-DEF‘thosehorses’

Asexpected, thepluralnumbermarker–anoccurs inside thedefinite

marker, indicating that the D where –e is realized is above the category ofnumber(see5a).Comparing(5a)to(3)wherethenounphraseisaprojectionofnumber(NumP)withDPbelowit,IarguethatNumPin(3)stillhasasecondDPlayeraboveitwheretheDisnotrealized(see5b).

This unpronounced D category above NumP (5b) provides double

definiteness to the noun phrase, a phenomenon which is common in CK andseveralotherlanguagesincludingHebrew,NorwegianandSwedish(seeDanon2008;Julien2003).

12.15-12.50Variational models of language acquisition: The best is still not goodenoughCallumHackett•NewcastleUniversity

The Variational Model of language acquisition developed in Yang(2002) is a Principles & Parameters system which describes acquisition as aperiod of competition between all and only possible adult grammars. If weaccept that a grammar can be roughly characterised as the combination of auniversal base (UG), a suite of specified parameter settings, and a lexicon oflanguage-specific sound-meaning correspondences, grammar competition caneasilybecomeanintractablelearningtaskifitfollowsthatchildrenaremeanttoselectbetweenfully-specifiedparametersets(Roberts&Holmberg,2005).TheVariational Model’s solution to this is to recast grammar competition ascompetitionbetweenparameterenumerations.

Ratherthaneachpossibleadultgrammarbeingfullyrepresentedinachild’s mind, children have access only to some small number of individualparameters,eachofwhichhasanassociatedprobabilityofbeingenumerated[+]or [–].Each timea childhearsorproducesanutterance, they ‘hypothesise’ anadultgrammarbycompilingalistofvaluedparameters,thereafterrewardingorpunishingparameterprobabilitiesuntiltheyhaveallconvergedonavalueof0or1(thusensuringthatallparametersettingsinanadultarefixedas[+]or[–]).

Legate & Yang (2007) showed that Variational Models can berigorouslyappliedtocross-linguisticcorpusanalysesof‘RootInfinitives’(verbsthat children produce in non-finite form in contexts where finiteness isobligatoryintheirtargetlanguages).Thisabilitytoestablishadirectconnection

42

8

that-3SFstole-3SF DEF-carUngrammaticality of sentence (3) gives substance to the Kiparskian

stance that the structural difference between factive and non-factivecomplement lies in subcategorization of the matrix verb as far as JA isconcerned. With the assumption that complements of factive verbs are, withupdateterminology,DPswithasilentnounanddeterminer(Schueler2016),thetwofactsoffactivecomplementsarestraightforwardlyaccountedfor.WiththefactthatDPsareabsoluteislandsinArabicgrammar(Mohammad1999;Soltan2007), the observation that no extraction whatsoever is allowed from factivecomplementswhichareembeddedunderDP follows.Furthermore, aDP layerthatdominatesCPfactivecomplementshasnosyntacticeffectsonA’-movementwithintheseclauses,somethingthatgivesrisetothepossibilityoffronting(andevenCLLD)insidesuchcomplements.

Asforobligatoryuseofthecomplementizerʔinn,(theobservationthathas been considered a problem to the DP approach as the complementizer issaidtobearealizationtothepotentialoperator;cf.Zubizaretta2001)IappealheretoapromisinglineofresearchbySchueler(2016)whoassumesthat it isthecomplementizerbutnotthelexicalverbthatisresponsiblefortriggeringthepresuppositionsassociatedwithfactivepredicates.14.25-14.50Calling Finnish speakers: Evidence for speaker and hearer syntax fromdiscourseparticlesAnnaHollingsworth•UniversityofCambridge

Discourse particles are oftendismissed into the fringes of linguistics.However, recent work on discourse-related elements has revived the ideaattributed to Ross (1970) of representing the notions of speaker and hearersyntactically, based on imperatives (Alcázar and Saltarelli, 2014), temporalrelations (Giorgi, 2010), and evidentiality and logophoricity (Speas, 2004),amongotherphenomena.Thistalkcontributestothisgrowingbodyofevidencethrough theanalysisofFinnishdiscourseparticles.Myaimsare twofold: Iwillshow that discourse particles occupy an important role in syntactic theory ingeneral,andarguethatthecurrentviewoftheFinnishleftperipheryhastoberefinedtoincludeadiscourse-relatedlayer.

Iwill focusontheparticles-hAn(amarkerof familiar information), -kO (the obligatory interrogative marker) and –pA (marking emphasis orexclamativeforce),aswellas–s(amarkerofinformalregister).Theparticles–hAn, -pA, and –kO, but not –s, are second-position clitics, always triggeringmovementtotheedgeofC.

The Finnish left periphery has typically been assumed to be rathersparseintermsofprojections.Standardanalysespositspec,CPasthehostforabroad range of elements, including second-position clitics, wh-phrases, andcontrastivetopicsandfoci(Vilkuna,1995).Huhmarniemi(2012)elaboratesthispicturebyarguingfortwodifferenttypesofmovementtriggersinspec,CP:theobligatoryfeature[force]forphraseshostingparticlesandtheoptionalfeature[focus] for contrastive elements. However, considerations pertaining to co-occurrence restrictions of the particles and their appearance in embeddedcontexts show that amore refined left periphery is needed to account for thedata.

projection of number (NumP) is aboveDP in the structure, I argue that thenounphraseproperstillhasaDPprojectionaboveit.

Basedon typologicaldata fromawide rangeof languages,Rijkhoff(2002) claims that referentiality elements such as the definite marker (D)takesscopeover,henceisstructurallyhigherthanquantityelementssuchasnumber(Num),asin(1).

Likewise,thestandardstructuralrelationbetweenNumPandDPis

thattheformerfallswiththescopeofthelatter, i.e.DPishigherthan NumP(see Ritter 1991, among many others). Also, according to the currentgenerativeapproach,DPisstructurallyparalleltoCPwhereDcorrespondstoC,andNumPisparalleltoTP.Thus, if it istruethatCPiscross-linguisticallylocated above TP in the clausal domain, DP should be above NumP in thenominaldomain,nottheotherwayround.

On the faceof theseproposals, thehierarchical structure in theCKnounphraseseemstobeacounterexamplewhereDPfallswithinthescopeofNum. The number marking is realized by the enclitic -an attaching to thedefinitemarker–ekewhich,inturn,attachestothenominalprojection(NP).

(2)esp-eke-an

horse-DEF-PL‘thehorses’This study adopts Chomsky’s (1995) Minimalist bottom-up

derivational theory. Thus, observing the example (2), I take the nominalprojection(NP)esp ‘horse’tomergefirstwiththecategoryDrealizedbythedefinitearticle–eke(see3),beforemovingtoSpecDP.TheDPisthenmergedwithNumrealizedby–an,andmovesinapied-pipingfashiontoSpecNumP,asshownbelow.

Given this structure, the category D where -eke is realized is

structurallylowerthanthefunctionalprojectionofnumber(Num),suggestingthat the CK noun phrase is hierarchically represented as NumP, not as DP.This contentionmilitates against all the standardassumptions raisedabove.However,IarguethattheNumPin(3)stillfallsunderanotherDPprojectionwheretheDcategoryisnullinconstructionssuchas(2),butmorphologicallyrealizedbythedefinitemarker–ewhendemonstrativesarepresent.Iprovideempirical evidence that the morpheme –e, which co-occurs with

41

9

First, the particles may co-occur on the same element, but two co-occurrencerestrictionscanbeobserved:theparticles–kOand–pA,ontheonehand, and – hAn and –s, on the other, cannot co-occur. Huhmarniemi (2012)argues that these restrictions arise from –kO and –pA expressing differentvaluesofthefeature[force],andsimilarly–hAnand–sencodingdifferentvaluesof the same feature. This suggests that the standardly postulated unitary CPmustbesplitfurthertoallowdifferentcategoriesofparticlestoco-occur.

Thispicturehastobeelaboratedfurtherbasedonthebehaviouroftheparticles in embedded contexts. I will adopt Haegeman’s (2006) typology ofembedded contexts, where different types of clauses are differentiated bywhether they encode Speaker Deixis (SD), a functional layer anchoring apropositiontoaspeaker.Whileperipheraladverbialclausesmirrorrootclausesin their structure in projecting SD, central adverbial clauses and clausesembeddedunderfactiveverbsdifferinmanifestingareducedstructurewithoutSD. Novel data show that the particles –hAn and -pA can occur in the formertypesofclauses,whiletheyareexcludedfromthelatter,reducedtype,providingevidenceforanSDlayerinFinnish.

Thedata show i) that theFinnish leftperiphery ismore refined thanstandardlyassumed,encodingaspeaker-andhearer-related layer,and ii) thatthe argument for discourse-related projections encoded in syntax iscrosslinguisticallycorroborated.

14.50-15.15ThesyntaxofsententialnegationinAbhaArabicdialectAliAlzayid•UniversityofEdinburgh

In this paper, I investigate sentential negation in the Abha Arabicdialect (henceforth AAD). It is a dialectal variation of Arabic spoken in thesouthwestofSaudiArabiaintheAsi:rregion.Tothebestofmyknowledge,Al-Azragi(1998)wasthefirstresearchertoexaminethesyntaxoftheAbhadialect.Itwasadescriptivestudyinnaturecoveringawidevarietyof topics includingnegation phenomenon. Based on her data, I argue thatAADhas four negationparticles: ‘maa’, ‘la’, ‘lim’, ‘lis’ as well as negative constructions in copularcontexts.Followingmanyworksadoptingwhatisknownas“Neg.Phypothesis”,I assume thatnegativeparticles inAADhead theirown functionalprojections.Moreover, the analyses of Benmamoun (2000) of Arabic functional categorieshavebeenadoptedwithslightmodificationstoaccountforthedatainquestion.15.15-15.40Doubletopicalisation,syntaxdoesconstrain!SherryYongChen•UniversityofOxford

Thispaperpresentsnewdatatoargueforsyntacticconstraintsontherelative order of nominal topics in double topicalisation constructions inMandarinChinese(henceforthChinese).

Chinese topics are derived via either base generation or movement(Badan&DelGobbo,2011;Huang,1982;Li,2000;Shyu,1995),andtheyalwaysappearinclause-initialposition.Beingatopic-prominentlanguage,Chineseisawindowintomanytypologicallyinterestingconstructions.Ofparticularinterest

(2)ʔal-bait-u li-r-radʒul-i (MSA)Def-house-NOM of-Def-man-Gen‘Theman’shouse’Saidthis,ħaggdiffersfromthelatterinthatitagreesinnumberand

genderwiththeprecedingDP,thepossessum(3a-b).(3)a.ʔas-sijjarah ħagg-at Mohammed (HA)Def-car.SG.F POSS-SG.F Mohammed‘Mohammed’scar’b.ʔas-sijjaraatħagg-aat Mohammed

Def-cars.PL.FPOSS-PL.F Mohammed ‘Mohammed’scars’

Whatseemsinterestingaboutħagg isthatithasshiftedfrombeinglexically used as a noun, meaning ‘right’, ‘righteousness’ or ‘truth’, to agrammaticalised functional item (a possessive preposition) denotingpossession.Moreover, thepresenceofanalyticgenitives, likeħagg, inArabicdialectshasbeenwidelytakenasa‘dialectalinnovation’(Harning1980).

Asforthesyntacticderivation,Iargue,followingDenDikken(2006),thatħaggcanbestructurallyderivedviaheadmovementofanullpreposition(Ro)headingtheRelatorPhrase(RP)toFooftheupperc-commandingLinkerPhrase(FP)formingacomplexhead,whereħaggisassumedtobeaPFformoftheamalgamatedhead(Ro+Fo).Thederivationrequiresfurthermovementto the Classifier Phrase (ClfP), situated immediately above the FP. Thismovement is necessitated by the involvement of the genitive preposition inpredicate inversion and in order to restore the original word order (DenDikken and Singhapreecha 2004). Additionally, the movement to classifierphrase is said to have an additional advantage of deriving preposition-possessum agreement by placing the genitive preposition in a Spec-headrelationwith thepossessum(DenDikken2006).Said this, Iargue further infavor of Ouhalla’s (2011) simplified version of the above in which theLinker+Classifier conflates into a single functional head, F[Agr] whichparallelsT[Agr]inclausesandcorrespondstoNuminDPdomain.ForOuhalla(2011), contra Den Dikken (2006), preposition-possessum agreement isassumedunderAgreerelationbetweenF[Agr]andthepossessumitcloselyc-commandsresultinginthevaluationofuφ-featuresofF[Agr]basedonthoseofthepossessum.Inthenextstepofthederivation,thePisraisedtoF[Agr]toget an inflected form of the preposition, i.e., ħagg, which agrees with thepossessum inφ-features. The final step comprises the possessum raising toSpec,F[Agr],yieldingtherightwordorder.11.50-12.15The noun phrase in central Kurdish: A projection of D (DP) not Num(NumP)RebwarTahir•NewcastleUniversity

This paper proposes that the noun phrase in Central Kurdish (CK,henceforth) is in factaprojectionof the functionalcategory(D), rather thannumber (Num). Although the surface form suggests that the functional

40

affected by literacy and grammar related activities, only a few studies haveattendedtothecontextandmethodofacquisitionofthebilinguallearners’L2.Indeed, it is not clearwhether the higher level ofmetalinguistic awarenessresponsible for bilinguals’ better performance in TLA is mainly due to thepreviousinstructed,formallearningprocessintheirL2orifitistheresultofthe accumulated experience in a naturalistic bilingual setting. The currentresearch aims to investigate the bilingualism specific variables affecting theprocess and outcomes of TLA, with a particular focus on the relationshipbetween the context and method of acquisition of bilinguals’ L2 (i.e.instructed/uninstructed; formal/informal) and the degree of metalinguisticawareness developed. It is worth looking at how the level of explicit andimplicit metalinguistic awareness of “primary” and “secondary” bilinguals(Hoffman, 1991) correlates with a higher level of proficiency attained in athird(oradditional)language.Inotherwords,isitthelevelofbilingualismorthe level linguistic knowledge in an L2 that plays a fundamental role tosucceedinTLA?(Bialystok&Barac,2014)

Theproposedquestionwillbeassessed through thecomparisonofthegeneralproficiencyattainedinL3bybilingualswhoseL2hasbeenaddedin a naturalistic setting and bilinguals whose L2 is the result of a formallearning process. First, a bio-data questionnaire will report the learners’previous linguistic history, motivation, and aptitude towards languages aswellas theirsocio-economicbackground.Second, twocloze tests in theirL2andL3willbeadministeredinordertoassesstheirdegreeofbilingualism(i.e.balanced/unbalanced) and their level of proficiency attained in L3. Third, aSelf-PacedReadingTaskconsistingof twoparts, i.e. a linearnon-cumulativeprocessing of semantically ambiguous sentences and a post-experimentquestionnairefocusingonthegrammaticalacceptabilityofthesentences,willbe used in order to provide independent measures of implicit and explicitmetalinguistic skills developed in their L2 and information about theparticipants’ cognitive abilities (Blom & Unsworth, 2010) and anticipationskills(Bonifaccietal,2011).

SESSION11 WEDNESDAY31STMAY,11.25-12.40

11.25-11.50OnthestructureofpossessiveħagginfreestatenominalsinHailiArabicEisaAlrasheedi•NewcastleUniversity

PossessioninHailiArabic(HA)canbeanalyticallyformedbytheuseofthepossessivepreposition,ħagg,whichintervenesbetweenthepossessumandpossessor,asin(1).

(1)ʔal-bait ħagg ʔar-radʒaal (HA)

Def-house POSS Def-man‘Theman’shouse’The element ħagg is similar to the preposition li in the Modern

StandardArabic(MSA),comparetheexamplein(1)with(2),below.

10 39

here is the so-called double topicalisation construction, where two nominaltopics are found in clause-initial position.What kind of surface structure cancapturethevariousinternalorganisationsofdoubletopicalisationconstruction?

To answer this question, I take the Minimalist Program as a broadtheoreticalframework,andadopttheTopicPhrase(TP)analysis(Gasde&Paul,1996)whichtreatsTPasa functionalcategorywithmaximalprojectionaboveanIPorarecursiveTP,withnominaltopicsoccupyingthespecifierposition ofTP,andtheTopicheadcanbeeitheremptyoroccupiedbytopicmarkers.Inmyanalysis, I first show that in double topicalisation construction, the base-generatedtopicmustprecedethemovedtopic:(1a)[Hua], (meigui-hua)itazuixihuan_____iFlower rose-flower3SGmostlike‘Flowers,roseshelikesthemthebest.’(1b)*[Meigui-hua]i,(hua),tazuixihuan_____irose-flowerflower3SGmostlike*‘Roses,helikesflowersbest.’

This constraint on the relative order of base-generated topics andmovedtopicsmaybeexplainedbythe localityconstraintonmovement,whichfollows the idea of minimal effort that places shortness requirement onmovementoperations(Li,2000;Hornstein,Nunes,&Grohmann,2005).

Secondly,whenbothtopicsarederivedviamovement,theymustbeina‘nested’dependencyrelationwiththeirgapsinthecommentclause:

(2a)[zhejianshi]i,(lixiansheng)jwogaosuguo_____j_____iThis-CLmatterMrLiItellPERF‘IhavetoldMrLiaboutthismatter.’(2b)*[lixiansheng]i(zhejianshi)jwogaosuguo_____i_____jMrLiThis-CLmatterItellPERF‘IhavetoldMrLiaboutthismatter.’

ThismaybeexplainedbythePathContainmentConstraint(Pesetsky,1982),whichrequiresthepathsofmovementtobeinacontainmentrelation.

These findings challenge the view that Chinese topics are merelyconstrainedviaasemantic“aboutness”relationwiththecommentclause(cf.Xu& Langendoen, 1985). They suggest that topicalisation is subject to syntacticconstraintsthatgoverntherelativeorderofdoubletopicsinChinese,whichfallunder the broad notion ofminimal effort. Chinese has a rigidword order butlittle inflectionalmorphology,which sometimesmakes it difficult to pin downthegrammatical functionofaconstituentbasedontraditionaldiagnostic tests.Nevertheless, the kind of syntactic constraints described above allow Chinesespeakers to interpret double topicalisation constructions in a relativelyeffortless manner. The grammar that underlies Chinese topicalisation isdesignedassuch,sothatminimaleffortisrequiredfromthespeakertoresolvepotentialambiguitiesdespitethelackofovertmorphologicalcues.

11

SESSION2 MONDAY29THMAY,15.50-17.0515.50-16.15Practicedlanguageattitudes:AcaseofmothertongueeducationinTaiwanChia-YingYang•UniversityofEdinburgh

Taiwan is a multilingual country where Taiwanese Mandarin is thenational language, with Taiwanese (Daighi), Hakka, and 10 Austronesianlanguages as generational or regional languages throughout the country.Moreover, the global Lingua Franca: English (see Seidlhofer, 2005; Graddol,2007; Abdullah and Chaudhary, 2012; Jenkins & Leung, 2013), also plays animportant role in the Taiwanese linguistic repertoire, as it is one of themainstream language subjects in the National Curriculum. Attitudes towardthese languagesconstituteawelldefined linguistichierarchy:Englishplacedatthetop,followedbyTaiwaneseMandarin,thenDaighi,HakkaandAustronesianlanguages(Hong,2002).Suchlinguistichierarchyhastriggeredalanguageshift:theyoungergenerations(theunder30s)ofwhicheverethnicgroupstheybelongto, have broadly switched to become monolingual in Taiwanese Mandarin(Census,2010).

ThisstudyfocusesonDaighiteachers,whoaremultilingualspeakersofat least three languages: Daighi, Taiwanese Mandarin and English, andinvestigatesthenotionofDaighiteachers’useof languageandtranslanguagingduring the interview and in the classrooms. As interviews are my main datacollectionmethod,Iinterviewed20teachers:10teachersteachinginthecapitalcity of Taiwan, Taipei, where based on the census 2010, intergenerationallanguageshiftfromDaighitoTaiwaneseMandarinisshowntobeatthefastestrate compared to the rest of the cities, as Taiwanese Mandarin is thepredominant language (see Yeh, Chan & Chen); and another 10 teachersteaching in Changhua,where the 2010 census indicates over 96% of residentreported using Daighi at home. I also observed two classes per teacher as asupportingdatacollectionmeans.

Arichsetoftranslanguagingdatawasfound.Inthispaper, I focusondiscussingwhatmaypotentially contribute to theoutcomeofonly4outof10DaighiteachersusingDaighiasthecodeofinterview.Ialsoexaminethefunctionofdifferentlanguageswhenusedininterviewsandclassroominstructions.Theresults show that language choices aredependent on the interlocutor, such asteachers’perceptionoftheirinterlocutor’sabilityinparticularlanguages,aswellas a form for the teachers to show their identity and professionalism.Translanguaging,ontheotherhand,isobservedtobeaformofcommunication,aspedagogy,asanactiontopromotelanguageequality,andaslanguageabilitythat teachers expect students to have to prepare them for the multilingualcommunity.Thispaperwilldrawonthedifferenttranslanguagingaspectsintheform of conversation analysis to show detailed language exchange, it alsodiscusseshowsuchtranslanguagingpracticeachieves thepurposesmentionedabove.

has mainly been limited to investigating groups of either younger or olderchildren, or specifically their early and late years of L2 exposure. However,comparingaccuracyanderrortypes intense-markingmorphologyproducedbybothyoungerandoldergroupsofL2-TDchildrenisessentialforrevealingageeffects in theirdevelopmentalpatternsof tensemorphology.Hence,ouraim is to provide a cross-sectional study, which investigates how tensemorphology develops among three different profiles of bilingual childrenacquiring English as L2 across a wide range of ages, and whether thedevelopmentaltrajectoriesoftheir3SG–sandpasttenseacquisitionprofilesareinlinewithpreviousstudiesthatsupportthepredictionofBybee’s(1995,2001) usage-based network model (Blom & Paradis, 2013; Paradis et al,2012).AgroupofWelsh-EnglishTDbilingualchildrenfrom7-9-yearsofage,ayoungergroupofL2-TDchildrenfrom4-6-yearsandanage-matchedgroupofL2-SLI peers were administered the tense probe from the Test of EarlyGrammatical Impairment (Rice & Wexler, 2001). Responses that had beentranscribed and scored on the TEGI were selected and coded in order toanalyzeindividualtensemorphologyelicitationprobes.Ouranalysisincludes1)ageeffectsofL2-TDchildrenontheiraccuracyanderrortypechangesovertime and 2) mixed linear regression to assess whether phonologicalsensitivity,wordfrequency,andL1transferhaveaninfluenceonperformanceof L2-TD children in the early or later stages of development, or acrossdevelopmental stages. Based on findings of howdevelopment patterns varybetweenyoungerandolderL2-TDchildren,asubstudywillcomparetheL2children with SLI to see how they develop tense morphology compared totheir age-matched TD controls. Results predict that tense morphology usecouldpotentiallyrevealdevelopmentalstagesasakey tounderstandingageeffects and profile differences across groups of bilingual children with andwithoutSLI.10.50-11.15Instructedvs.uninstructedbilinguals:TheroleplayedbymetalinguisticawarenessinthirdlanguageacquisitionFrancescaD’Angelo•UniversityofEdinburgh

The present work explores the relatively new field of research

investigatingThirdLanguageAcquisition(TLA),whichconnectstwodomainsthathave traditionallyoverlookedone another: second language acquisitionandbilingualism.

The expansion of research intomultilingualism in the past decadehas been the result of a growing awareness that acquiring a third language(L3) isadifferent,morecomplex,andmultifacetedprocesscomparedto theacquisition of a second language (L2). Numerous studies on the positiveeffects of bilingualism on TLA relate the advantages evident in bilinguallearners to the influence of bilingualism on cognitive development and,specifically, to metalinguistic awareness (MLA) and communicative skills(Cummins, 1978; Thomas 1988; Cenoz & Genesee, 1998; Jessner, 1999;Jessner,2006;Jaensch,2009).Thishasbeenconsideredasanindirecteffectofbilingualism,that is, “bilingualismaffectscognition,MLAandcommunicativeskillsandthesefactors,inturn,affectTLA”(Cenoz,2003).

Although it is has beenwidely acknowledged thatMLA is strongly

38

Dependentontype, languagescouldallowa fusionofgrammaticalmeaningsin single morphemes (semantic complexity), draw on free and boundmorphemes (syntactic category) to various extents, and exhibit variants ofmorphemes based on their phonological environment (morphophonologicalregularity). Differing levels of these predictors in morphemes have had astatisticallysignificanteffectonthesuccessoflearningthemorphemesinthestudies.

Inthefirststudy,conductedwithnativespeakersofthreedifferentlanguages,themothertongueofthelearnershowedaneffect,butislikelytobe correlatedwith cultural factors.The language typeof theL2, in contrast,proved to be a reliable predictor of learning difficulty, even independent ofthe learner’s mother tongue. Further, the isolating artificial language,exhibitinga largeamountof freemorphemes,whichwerealsonon-complexintheirstructure,waslearnedsignificantlybetterthantheothertwotypes.Atanearlystageoflearning,semanticcomplexityappearedtoinhibitsuccessofacquisitionlessthanmultimorphemicwordswithanunderlyinglesscomplexstructure. These findings emphasize the role of visually salient freemorphemes and they allow for the careful assumption that facilitatedwordsegmentation due to a low level of semantic complexity did not contributemuchtoeaseoflearning.

In the second study, conducted with English native speakers,however, low semantic complexity did facilitate ease of learning incomparison to a conditionwith semanticallymore complexwords.Due to adifference in word length between both studies, the findings suggest theexistence of a threshold at which clarity of word structure does no longercounterbalancethedifficultyresultingfromwordlength.Overall,thefindingsfrom both studies show that certain word structures are easier to acquirethan others, independent of the learner’s mother tongue, and consequentlysupporttypologicallybasedaccountsoflearningdifficulty.

SESSION10 WEDNESDAY31STMAY,10.25-11.1510.25-10.50The development of tense morphology by Welsh-English bilingualchildrenwithandwithoutSpecificLanguageImpairment(SLI)HyowonKwon&VickyChondrogianni•UniversityofEdinburgh Todate,muchresearchconcerningtheacquisitionofEnglishtense-marking morphology as a clinical marker has focused on comparingmonolingual childrenwith SLI to theirTDpeers (Leonard, 1998;Rice et al.,2000;Rice&Wexler,2001;Bishopetal.,2006).StudiescontrastingchildrenwithandwithoutSLIacquiringEnglishasasecondlanguage(L2)haveshownthatL2childrenhavemoredifficultieswith tensemorphology inproductionthanmonolinguals,andEnglishL2childrenwithSLIshowexceptionaldeficitsin tense morphology compared to their L2-TD peers (Chondrogianni &Marinis, 2011; Paradis, 2005; Paradis et al., 2008). Therefore, looking ataccuracy and error types of tense morphology by L2 children with andwithoutSLIcouldfunctionasthepotentialclinicalmarkerthatdistinguishesTDfromSLIinEnglishL2children(Blom&Paradis,2013;Paradis,2008).Toourknowledge,researchtappingintobilingualchildrenwithandwithoutSLI

12 37

16.15-16.40Emerging self-identities and emotions: A qualitative case study of tenSaudistudentsintheUKOunAlmesaar•UniversityofEssex

ThistalkaimstouncovertheprocessesofL2identityconstructionanddevelopmentamongSaudiArabianstudents in theUnitedKingdom.Guidedbythetheoryofpossibleselves (Markus&Nurius,1986), thetheoriesofsituatedlearning(Lave&Wenger,1991)andimaginedcommunities(Norton,2000),thispresentation will examine the participants’ past and present learningexperiencesandemotions,andtheireffectsonparticipants’orientationandself-perception to English writing. Dörnyei’s (2005) notion of the Ideal L2 self isutilised toexplore identity andSaudi students’ viewsof themselvesaswritersovertimeandinvariouscontexts.Thisexploratorycasestudyemployedvariousinstrumentstocollectdata includingasurvey, twosemi-structured interviews,journals, think aloud protocols, and writing samples. Thematic and narrativeanalyseswereutilisedtoanalysethedata. The preliminary findings show that all participants had a negativelearningexperienceinSaudiArabia;theywereunmotivatedandlackedchancestopracticeEnglishparticularly inwritten form.The learningexperiencehadalittleeffectonsomeparticipantsandnoeffectonothers in theconstructionofpossibleselves.IntheUK,allparticipantshadapositivelearningexperienceandtheirlanguageskillsimprovedsignificantly.TheirL2learningexperienceplayedahugerole intheconstructionofpossibleselvesaswellas inreactivatingandcreating desires in becoming better and more successful writers for mostparticipants. It also helped in sustaining and reinforcing participants’ L2possible selves through creating chances for the participants to be active andautonomous learners. Furthermore, participants’ future possible selves wereperpetuated via joining variousEnglish language courses,master courses, andspendingalongtimeintheUK.TheirviewsofthemselveskeptonshiftingandchangingduetotheiracademiccircumstancesandemotionswhichaffectedthewayparticipantsviewedthemselvesaslanguageusersandEnglishwriters.16.40-17.05Metadiscourseintheclassroom:Acomparativeanalysisofnativeandnon-nativeEFLteachersXinxinWu•UniversityofAberdeen Metadiscourse, as a rhetorical and pragmatic strategy used toconstruct discourse and express the speaker’s or writer’s attitude, has greatpedagogical implications for the classroom teaching of English as a foreignlanguage (EFL) teachers, especially for the non-native EFL teachers whoseclassroomdiscourseformsthemediumaswellastheinputfortheEFLstudents.Researchintometadiscourse,however,hasbeenprimarily focusedonthefieldofwrittendiscoursesuchasdiscourseanalysis,pragmaticstudiesandlanguageeducation,whilethespokenacademicEnglish,especiallycomparativestudiesinteachers’ classroomdiscourse has remained largely unexplored. Based on twocorpora of EFL teachers’ English for general academic purposes (EGAP)classroomdiscourses:5EGAPclassesfromUKuniversitiesand5EGAPclassesfromChineseuniversities,thisexploratorystudyattemptstoinvestigatetheuse

thecontext(thestateoftheworlditself).Perspectiveandcontextinteracttoproduce the agent’s mental state, which influences how likely they are tochoose the different potential referents as their communicative intention.Whiletheagent’sperspective is fixed, thecontext ischangeable;renderingadifferentmentalstate foreach interaction. Inadditiontoaperspective,eachagenthasalexicon(mappingofsignalstoreferents),whichtogetherwiththecommunicativeintentionproducesutterances.

Duringdevelopment,eachnewagentobservesasetofcontextsandcorresponding utterances from their cultural parent. From these twoobservable variables the learner has to infer two unobservable variablessimultaneously: the parent’s lexicon and perspective, using Bayesianinference.Thedevelopmentalresultsshowthatthisjointinferencetaskleadsto co-development. A parent’s lexicon can only be inferred correctly if thelearner can represent the possibility of the parent having a differentperspectivefromtheirown,andtheperspectivecanonlybeinferredcorrectlyif the parent uses an informative language. Thus, learning of lexicon andperspectiveareinterdependentinthismodel.

This leads to the question of under which circumstances apopulation of such agents is able to establish a conventional language fromscratch,i.e.ifallagentsinthefirstgenerationuseacompletelyuninformativelexicon. The evolutionary results show thatwithout any additional selectivepressure, an informative language does not evolve. Selection for successfulcommunication can provide the necessary pressure to lead populationstowards convergence on an optimal language. Interestingly however,selectiononperspective-takingcan leadpopulations to thebeginningsofaninformative system as well; stopping at whichever point the language isinformative enough to learn perspectives correctly given the number ofobservationsthatthelearnersreceivefromtheirculturalparent.Thismodelthusgivesinsightintotheeffectsofanindividual-levelinteractionofcognitivecapacitiesonpopulation-widedynamicssuchasestablishingandmaintainingastablesignallingsystem;therebyconnectingproximateandultimatecausesoflanguageevolution.

09.50-10.15Investigating effects of morphological types and their word formationpatternsinSLASvenjaWagner•UniversityofEdinburgh Previous research in SLA has shown that knowledge of structuresthroughthemothertonguecansupporttheacquisitionofsimilarstructuresinan L2 (Gass & Selinker, 2008). At the same time, factors such as semanticcomplexity and morphophonological regularity have been shown to impactacquisition ease also independent from the learner’s mother tongue(Goldschneider & DeKeyser, 2001). My research combines both theoreticalapproaches and investigates to which extent morphological language typescan predict learners’ success, both in relation to and independent from thetypeofthelearner’smothertongue.Intwoartificiallanguagelearningstudies,I have modeled three morphological types, isolating, inflectional, andagglutinating, based on their mapping of grammatical meaning ontomorphemes and the combination of morphemes to form complex words.

36 13

ofmetadiscourseinEGAPclassesbyEFLteachersacrossChinaandUK.Inlightof Hyland’s (2005) interpersonal model and Adel's (2010) reflexivemodel ofmetadiscourse, a revised model is to be proposed, and the two corpora arecompared to analyse the differences between their metadiscourse uses ingeneral,andindifferentmodesofclassroomteachingactivitiesbasedonWalsh(2006, 2011, 2013) in particular, and to explore the influence of pedagogicalcontent and context on teachers’ choice of metadiscourse in classrooms. Thisstudy also has further implication on EFL teachers’ training and developmentprogram. SESSION3 MONDAY29THMAY,17.15-18.3017.15-17.40Across-linguisticapproachtoadpositionsinthirdlanguageacquisitionSakineÇabuk•MiddleEastTechnicalUniversity This study explores the role of cross-linguistic influence in thirdlanguageacquisitionprocessbyexaminingEnglishadpositions.Comprehension,processing and production of English prepositions (in, on, at, behind, over, to)were examined through off-line and on-line data collection tasks to find outwhich of the two known languages (L1 or L2) is the major source of cross-linguisticinfluenceontheacquisitionofEnglish(L3)adpositionsgiventhefactthat adpositions are morphologically and syntactically different in Turkish,Kurdish and English languages. The main reason behind the choice of theseparticular prepositions lies in their morpho-syntactic properties in Kurdish,Turkish and English. While some adpositions have similar representations inthese languages (e.g., behind, overappear as prepositions in bothKurdish andEnglish), some others have different representations (e.g., in, on, at appear asprepositions in English, case markers and/or postpositions in Turkish andprepositionand/orcircumpositionsinKurdish).ParticipantsofthestudywereTurkish-Kurdish bilinguals who formed experimental group and L1-Turkishmonolinguals who served as control group. Two off-line picture descriptiontasks (picture description task with multiple choices and teddy bear picturedescription task) and an on-line self-paced reading task were employed tocollect data. The finding of the study revealed that Turkish-Kurdish bilingualswere better in comprehending, producing and processing target prepositionsthanL1-Turkishcontrolgroup,particularlywhentheyhavestructuraloverlapsbetween the adpositional systems of L1-Kurdish and L3-English. The findingsaresuggestiveoftypologyasanoverridingfactorincross-linguisticinfluenceintheacquisitionofL3English.StructuraloverlapsbetweenKurdishandEnglishfacilitatedtheacquisitionofEnglishprepositionforTurkish-Kurdishbilinguals17.40-18.05Think aloud protocols: A window into the minds of how English as aforeignlanguagestudentsprocessfeedbackSharifaAlHarrasi•UniversityofStirling Therehaslongbeenadebateonwhetherwrittencorrectivefeedback

35

construction in evolutionary theory, this paper tries to critically assess towhat extent such correlational claims can be meaningfully integrated incurrent accounts of language evolution andwhether LA should be added asone of themajor driving forces underlying linguistic structure. In line withRoberts & Winters (2013), I argue that correlational studies like the onesabovecangeneratehypotheses,butshouldbeviewedwithcaution,sincetheycannotprovidecausalexplanations.Inordertoaddresstheroleoflanguage-external pressures, I suggest to a) utilize such large-scalestatistical/correlational studies to formulate questions about possibleexplanationsfor linguisticdiversityandb)test thesehypotheses incarefullycontrolled laboratory experiments. These allow isolating mechanismsresulting in structural patterns that resemble distributions we can observeamongthelanguagesoftheworld.AsanexampleforhowtheorieslikeLAcanbenefit from the integration of complementary empirical approaches, I willpresent data from two experiments. The first one had dyads solve acoordinativemazegame,wheremanipulationoftheshared-taskenvironmentcould motivate different linguistic strategies based on differentconceptualisations of the otherwise identical task. The second experiment,relying on the iterated learning paradigm (Kirby, Griffiths & Smith, 2014),investigatedhowusagecontextcaninfluencetheemergenceandevolutionofoverspecificationofasemanticmarkerinanartificiallanguage(Tinits,Nolle&Hartmann, 2017, in press). In sum, these studies show how hypothesesconcerningLAcanbeformulatedandtestedbyintegratingcorrelationalandexperimentalwork.

09.25-09.50Theculturalco-evolutionoflanguageandmindreadingMarieke Woensdregt, Simon Kirby, Chris Cummins, & Kenny Smith •UniversityofEdinburgh Language use requires an understanding of communicativeintentions,whichinturnreliesontheabilitytorepresentother’sminds(e.g.Moore,2016).Thedevelopmentof suchmindreading skills formeda crucialstep in hominin evolution, because it allowed for the expression andrecognition of communicative intentions, thereby paving the way for thecooperative information sharing we find in humans today. The ability torecogniseandinfercommunicativeintentionsalsoplaysanimportantroleinlanguagedevelopment,asevidencedbystudiescorrelatingmindreadingskillsand word learning (e.g. Parish-Morris, Hennon, Hirsh-Pasek, Golinkoff, &Tager-Flusberg,2007).Thisrelationshipbetweenlan-guageandmindreadingmay be reciprocal; the acquisition of language has been shown to unlockfurtherlevelsofmindreadingdevelopmentintheindividual(e.g.Lohmann&Tomasello, 2003; Pyers & Senghas, 2009). Furthermore, Heyes and Frith(2014)arguethatthesophisticatedmindreadingskillsofmodern-dayhumansareinpartaresultofcumulativeculturalevolutionwhichwaslikelyaidedbylanguage.

Inthispaperwepresentanagent-basedmodelthatinvestigatestheimplica- tions of such a bidirectional interaction between language andmindreading.Inthismodel,mentalstatesareimplementedastheproductofaninteractionbetweenanagent’s‘perspective’(theirviewontheworld)and

14

(WCF) facilitates learners’ development in a second language (L2), and, if so,whethermoreorlessexplicittypesofWCFaremoreeffective.Researchersarekeen todiscoverhow learnersprocessWCFand tounderstandhow,whyandwhendifferentstrategiesofWCFfacilitateL2development. The research project described in this talk investigates theeffectiveness of two types of WCF (direct and indirect) in improvinggrammatical accuracy of 77 Omani EFL students on two linguistic structures(comparativesandprepositionsofspace)bymeansofpre-test,immediateposttestanddelayedposttest. Thinkaloudprotocolswerealsousedwith6studentsfromeachgroupto investigate how students process the feedback given (direct vs. indirect)whiledoing revisionsandnew tasks.Thispresentationwill focuson the thinkaloudprotocols.Drawingondatafromthestudy,Iwillshowthatbothcognitivefactors and my own scaffolding approach influenced the way that learnersprocessedtheWCF.Furthermore,students'reportsrevealedinterestingfindingsrelatingtohowdifferentindividualsprocessedtheWCF. Findingsshowthatgreateraccuracyonpost-testsresultedwhenWCFwasnegotiated.ThisresultchimeswithNassaji'sstudy(2012)whichfoundthatnegotiated feedback ledtogreateraccuracyonpost tests thannon-negotiatedandminimallynegotiatedfeedbackfortheuseofarticles.18.05-18.30TheinfluenceoforalreadingonKoreanlanguageproficiency:AcasestudyJihyeEo•YonseiUniversity Thisstudywasdesignedtoinvestigatetheeffectoforalreadingon26collegestudentsofKoreanasaforeignlanguage.Traditionally,oralreadinghasbeen known teaching methods to teach a foreign language, especially fordevelopingreadingfluency.Thesedays,oralreadingthatwentunnoticedintheprevious researches is now being revalued as a significant skill to improveforeign language learners’ communicative competence.Alongwith this changeto new awareness and evaluation of oral reading, there have been severalresearchesthatstudytherelationshipbetweenoralreadingandotherlanguageskills. ThestudywasconductedfromtwodifferentKoreanlanguagecoursesataprivateuniversityintheUnitedStatesin2014and2015.Intheselanguagecourses,whereheritageandnon-heritagelearnersofKoreanlearnedthetargetlanguage together, total14 students enrolled in “IntermediateKorean” and12students enrolled in another “IntermediateKorean”. The researcher itselfwastheinstructorofthesetwoclasses,whichconsistedoffour50-minutemeetingsforthedurationofa15-weekssemester.Thestudentswereaskedtoreadaloud26passagesandrecord themselvesovera13weekperiod.Eachrecodingwasreviewedandcommentedbytheresearcherwhothenratedeachrecordingonanumerical scale. Five perspectiveswere used to reviewed and commented forthe participants; (1) accuracy of the word pronunciation, (2) expression andvolume, (3) phrasing, (4) pace and (5) prosody, and all performance weregradedona1 to4 scalewith4being thehighest.And the testof readingandspeaking testwas carried out oneweek after the performance of oral readingwasconducted.

Theresults in thisstudyshowedthatoralreadingeffectsonnotonly

35

34 15

learner’s proficiency and but also reading proficiency. Findings first indicatedthat participants with lower reading fluency improved their readingcomprehensionthroughthisexperiment.Second,it showedthatoralreadingiseffectiveinpromotingautomaticphonologicalcodingandacceleratingthespeedofvocalspeech.Theresultssupportapedagogicalimplicationthatoralreadinghas an instructional value to improve learners’ speaking and pronunciationalfluencyaswellasreadingfluency.SESSION4 TUESDAY30THMAY,09.00-10.15 9.00-9.25ScarystatisticsandhistoricallinguisticsDaisySmith•UniversityofEdinburgh

Theavailabilityof large-scalediachroniccorporahasmadeitpossibleto investigate historical linguistic phenomena using sophisticated statisticalmethodology. Gries (2015, 97) describes corpus data in general as“observational and, thus, usually unbalanced andmessy/noisy”. He points outthat techniques frequently employed in other areas of linguistics, particularlythose which use experimental methods, can be usefully applied to corpuslinguistics. More than that, he suggests that these methods are activelynecessitatedbytheverynatureofcorpusdata.

Historicalcorpusdata isnoexceptiontothisgeneralisation,butHistoricalLinguistics is a field inwhich the scope for ‘bigdata’ analysishasbeenunder-standablysmallerthaninothers.Compare,forexample,thefollowingcorpora:

1. TheEdinburghTwitterCorpus(Petrovic,Osborne,&Lavrenko,2010):acorpusofsocialmediainteractionscontaining97milliontweetswithacom-binedtotalofover2billionwords.

2. ALinguisticAtlasofEarlyMiddleEnglish(LAEME)(Laing,2013):acor-pus of Middle English (ME) texts containing approximately 300transcribed Medieval manuscripts with a combined total ofapproximately650,000words.

Petrovic et al.’s corpuswas extracted automatically over a period of

twomonths using Twitter’s streaming API. By contrast, LAEMEwas compiledmanually over the course of 20 years by the manual sourcing, reading,decipheringandtranscribingofmanuscripts(Laing,2013).Evenconsideringtherelative length of the LAEME texts compared with 140-character-maximumtweets, these two corpora are clearly on vastly different scales. Where theelectronic data from Twitter can be extracted automatically, the LAEME data,and all similar historical corpus data, must be painstakingly transcribed anddigitised.

How,then,inafieldwherehand-drawnisoglossesareformanyanot-so- distant memory, should a researcher embarking on a complex statisticalanalysispresenttheirresultstothehistoricallinguisticcommunity?

In this paper, I present an analysis of data fromA Linguistic Atlas ofOlder Scots (LAOS), firstly considering what information can be gleaned from

at least the 1950s (Björseth, 1958). It might therefore be expected to bepresent inolderGothenburg speechaswell. InUppsala,Viby-ihasnotbeenrecorded,butitisstillpossiblethatithasfounditswaytherefromStockholm,whichisonlyanhour'sdriveaway.Uppsala ishoweverregardedasclosetothenationalstandarddialect(Källskogetal,1993),whichmeansthatitmightalsoprovideagoodreferencepointforstandard[iː].

The aim of the paper is thus to investigate how far along the shifttowardsViby-ihascome,primarilybylookingatageandlocation.However,IamalsointerestedinthedistributionofViby-iacrossothersocialfactors,suchasgenderandsocialclass,sincethismighttellusbothaboutitsspreadanditssocial value. In urban areas in particular, Viby-i is associated with highprestige(Bruce,2010),whichmaymakeitmorecommonincertaingroups.Itisalsopossiblethatspeakersproducethevoweldifferentlybasedonsocialorgeographicalfactors.

Theresearchispartofalargerprojectinvestigatingtherelationshipbetweenarticulation,acousticsandsociophoneticsintheproductionofViby-i.Thesubsamplepresentedhereconsistsof18femalespeakersaged20-25and50-80producingphrase-listutterances,withapproximately50/iː/tokensperperson. The remaining long Swedish vowels /yː, ʉː, eː, øː, ɛː, ɑː, ɔː, uː/ areincluded for context. The acoustic analysis is primarily based on formantvaluesforF1-F3.

SESSION9 WEDNESDAY31STMAY,9.00-10.1509.00-09.25Linguisticnicheoracademicniche?Theroleoflanguage-externalbiasesinlanguageevolutionJonasNölle•UniversityofEdinburgh Over the past decade, a number of major claims have been madeabout the fundamental nature of human language(s). For instance, from anevolutionaryperspective,ithasbeenarguedthatthestructureoflanguagesisnot theoutcomeofasinglebiologicaladaptation,assomewouldargue(e.g.,Berwick & Chomsky, 2017), but rather the result of biological-culturalinteraction (e.g. Christiansen & Chater, 2008; Kirby 2016), which – overhistoricaltimescales–gaverisetostructureanddiversity(Evans&Levinson,2009). On such a view, language is a complex adaptive system (CAS, seeBeckneretal.,2009)thatconstantlyevolvesasitislearnedandusedwithinapopulationofagents.Morerecently,thisviewhasbeencomplementedbytheidea of linguistic adaptation (herafter LA, see Lupyan & Dale, 2016 for areview). The hypothesis is that languages as CAS tend to adapt to theirphysical,socialortechnologicalenvironment,similarlytoorganismsadaptingto their local ecological niche. This has been supported by a number ofcorrelations that have been found in recent years, e.g., the relationshipsbetween morphological complexity and population size (Lupyan & Dale,2010),linguistictoneandhumidity(Everettetal.,2015),lexiconandclimate(Brown & Lindsey, 2004; Regier, Carstensen & Kemp, 2016) as well asanatomicalandgeneticbiases(Dediu&Moisik,2016;Dediu&Ladd,2007).

Inspired by Gupta and colleagues’ (2017) critique of niche

some basic descriptive statistics. Secondly, I present the results of ageneralised linear model (GLM) fit to the data, showing how this canprovidefurtherinsightintotrendscomparedwithbasicstatisticalmethods.Lastly,Ipresentageneralisedadditivemodel(GAM)fittothesamedataset,againshowinghowthismethodimprovesinsightintothedata.

9.25-9.50Language, identity and community: The use of third-person neuterpossessivesofearlymodernpreachersHiroshiYadomi•UniversityofGlasgow This paper examines the use of third-person neuter singularpossessive pronouns (its, his, of it, and thereof) of 20 Early Modernpreachers. In theperiodexplored, thenewformitswasnewly introducedand other variants were also available to express the same meaning(Nevalainen & Raumolin-Brunberg 1994; Barber 1997; and Lass 1999). Iwillshowthatthereisahugevariationinthechoiceofthird-personneuterpossessives in preachers’ language and discuss possible factorsconditioningthevariationbetweenindividuals.Inadditiontolinguisticandmicrosociolinguistic factors (sex,ageandclass), Iwill shed lightonmoremicro sociolinguistic parameter of sermonwriters, namely their religiousidentity.ThisisapartofmyPhDprojectanalysingthecorrelationbetweenEarlyModernpreachers’religiousidentityandtheirlanguagepractices.Theproject aims to describe how individuals and communities interact withongoinglanguagechanges. The data used in this study is retrieved from the Corpus ofSermons in Early Modern England, which has been compiled to analyselinguistic variation between individual speakers in the sole genre ofsermons (Yadomi 2016). The corpus contains one millions words, with50,000 words from sermons ascribed to each of the 20 preachers. Thesermons included in the corpuswere published between 1588 and 1660,duringwhich the third-person neuter possessive pronouns experienced ahuge,rapidshift. The study will focus on the relationship between preachers’languageuse and their religious identity, towhicha community isdeeplyrelated.EarlyModernpreachersformadiscoursecommunitywhichshares“common goals” and “linguistic practices” (cf. Swales 1990) as well asprofessionalidentity.However,thecommunityisbynomeanshomogenousin that therewas a rather clear divide amongmembers in terms of their religious identity, which lies subordinate to the professional identity(Kiesling 2013). In any sort of community, there arises voluntary orcompulsory social force which accommodates its members to thecommunitynorm.TheconfessionalstatusofEarlyModernEnglandcanbeclassified to Anglican conformist, moderate Puritanism and radicalPuritanism(Lake1988:7). In the presentation, I will discuss how ideologically conflictingparties within the discourse community had different preference inlanguage choice. Pilot studies show that Puritan preachers tend to retain

28

33

Twitter hashtag use and the rhematization process it undergoes. UsingTwitter extracts as the primary data, this paper first looks at how hyper-American concepts index ‘Merica,’ then goes on to assert that ‘Merica’ hasdeveloped a natural relationship as an icon for the heightened, polarizingcliches of American society. With reflexivity being at the heart of indexicalinterpretation,theimportanceofcontextisreinforcedinthispaper.Lifestyleindexes provide the context needed for other ‘Merica’ indexes to take on areflexiveform.Withsuchacomplex,richfeature,thisanalysisemphasizesthescale of reflexivity seen among indexes. This paper suggests that speakersdrawon‘Merica’whenitsupportstheirpoliticalorsocialcritiqueinTwitterpublications. With its emphasis on indexicality, this paper calls for furtherexplorationintohowothersocialmeaningislinguisticallygivento‘Merica.’ 16.45-17.10ThechangingSwedish/iː/vowel:Evidence fromthreeCentralSwedishcitiesFabienneWesterberg•UniversityofGlasgow

TheCentralSwedish/iː/vowelistraditionallyrealised[iː],buttoan

increasing extent, speakers are beginning to use another variant known asViby-i. This is a vowel with an unusual “thick”, “dark”, “buzzing” quality(Borgström,1913)foundin“severalscattereddialects[acrossSweden],bothin rural areas and in the city dialects of Stockholm and … Gothenburg”(Björsten & Engstrand, 1999). Viby-i has not been extensively studied, andthere aremany questions surrounding e.g. how its unusual vowel quality iscreated,whetheritisproduceddifferentlyindifferentlocations,orevenhowfaracrossSwedenithasspread.

This paper presents an acoustic apparent-time analysis of a smallsample of speakers from Stockholm, Gothenburg and Uppsala in CentralSweden (Fig 1). Only the first two cities are known to have Viby-i. InStockholmitseemstobearelativelynewfeature,whichwasfirstdocumentedinthe1990s(Kotsinas,1994),andthenprimarilyusedbyyoungupper-classfemales.Sincethisgroupisoftenattheforefrontoflanguagechange(Trudgill,1974), it is likelythatthefeaturehascontinuedtospreadwithinStockholm,but that it is most common in young and middle-aged speakers. InGothenburg,Viby-i isconsiderablyolder,withearlyrecordsdatingitbackto

16

scientificdiscipline:thecontributiontosocialtheory.In this talk, I will engage in the metatheoretical discussion by

rethinkingthiscontributiontosocialtheoryfromanotherperspective.First,Iwill philosophically reframe the three most prominent subdisciplines ofsociolinguistics, namely linguistic variationism, linguistic anthropology andsociologyoflanguage.Iwillfocusontheformertwoastheseareconsideredto be antagonists in contemporary metatheoretical debate. Subsequently, Iwillshowhowsociolinguisticinterestshavechangedrecentlybypresentingashort content analysis of contemporary sociolinguistic scholarship, andwillargue that the above-mentioned distinction of sociolinguistic subdisciplinesno longerdoes justice to the complexityof sociolinguistic explananda. Iwillbriefly summarise how the contributors to the Coupland (2016) volumereflectonthischange,andwillgoontoarguethattheirratherprogrammaticstance can be underpinned by drawing from the scientific realism/meta-inductive pessimism debate within the philosophy of science (cf. Ladyman2016).Iproposethatfromtheviewpointofascientificrealist,thetransitionawayfromclassicalvariationismtowardsanall-encompassingsocio-semiotictheory of language is a sign of a maturing science. This maturation can beexplained by the grade of self-awareness of value-laden presumptions andexplanatory value, which both depend on the degree of ontological self-justification.16.20-16.45TheUnitedStatesof‘Merica:IndexicalityandTwitterHashtagUseRachelMoyer•UniversityofEdinburgh InaheatedtimeintheUnitedStates,socialmediaandmicrobloggingplatforms, likeTwitter,havebeenusedasforumstoexpresspoliticalbeliefs,social critiques and frustrations deriving from the cultural constructs inAmericansociety.Exploringthemediaplatformswherepeopleexpresstheirexasperation for thecurrent social climate, cangive insight intohowpeopleindex the society inwhich they live. Recent sociolinguistic research (Shapp2014; Page 2012; Papacharissi 2012; Chang 2010) has discussed how thesocial has manifested in Twitter hashtag use, finding that hashtags havedeveloped past their original organizational function and now contributeadditional meaning to the message. Hashtags can be used in many socialprocesses like self-evaluation, image-building and self-presentation. Newstudieslinkingresearchbetweenthesociolinguisticsofmicrobloggingsitestoindexicaltheoryarestillcalledfor.

Indexical research has discussed the importance of context inattributing meaning to a specific linguistic norm (Jaffe 2016; Eckert 1989,2008;Gal1978,Silverstein2003).Byintroducingtheideaofindexicalorders(Silverstein 2003), macrosocial theories meet microsocial practice.Indexicality emphasizes that the relationship among indexes is a cyclicalprocess of reinterpretation (Eckert 2008, 463). Explicating the network ofindexesinanindexicalfieldallowsindexestobeinterpretedinrelationtooneanother. Using indexes to create a context for hashtags, indexicality meetssocialmediainthispaper.

The#MericatrendonTwitterismostexemplaryofthepoliticalandsocialschismsintheUS.Thispaperanalyzestheindexicalfieldof‘Merica’in

older forms for third-person singular possessive (particularly thereof). Thelinguistic evidence may illustrate the strong conformity of Puritans to theWordsofGod,thetraditionofEnglishvernacularBibleswhichretainarchaicformsoflanguage.Thesamecorrelationhasbeenobservedintheuseofthird-person singular inflection (-th vs. -s)which have been reported in previousconference presentations; thus, I will attempt to generalize the overallrelationshipbetweenreligiousidentityandlanguagepractice.9.50-10.15DiscourseandtopicalizationinoldEnglishsubordinationSergioLópezMartínez•UniversityofOviedo Thecloserelationbetweendiscourseandsyntax inOldEnglishhasbeen a favourite topic for research over the last few years, as seen, forexample, in Kemenade and Los (2006), Kemenade (2009), or Kemenade &Milicev(2011).However,mostoftheexistentworkontheinterplaybetweeninformation structure and syntax inOld English focuses onmain sentences,andtheinstancesinwhichsubordinatesentencesarestudied,itisinrelationtothediscourse-relatedparticlesþaandþonne. Inpreviouswork(LopezMartınez2016),IprovideddatasupportingtheclaimthatembeddedtopicalizationispossibleinOldEnglishsubordinatesentences, and it was suggested that this phenomenon could be related todiscourse factors, in connection to van Bergen (2015). Thus, the aim of thepresentpaper is tostudyhowdiscourse factorsmay influencetopicalizationinsubordinateclauses.Inordertodoso,alargecorpusofprosetextsfromtheOE period will be analysed, examining not only the instances of embeddedtopicalization insubordinatesentences,butalsohowdiscourseaffects thosetopicalizedelements.Thisstudywillpayattentiontowhetherthoseelementsarethematic(i.e.giveninformation)orrhematic(newinformation),analysingthe discourse preceding the relevant subordinate clauses in relation to thecurrenttheoriesonInformationStructure.SESSION5 TUESDAY30THMAY,10.25-12.0510.25-10.50Mechanisms of cognitive control in bilinguals: The empiricalgeneralizabilityofthe‘bilingualadvantage’MichelaBonfieni•UniversityofEdinburgh The relation between the bilingual linguistic experience andcognitive control is hotly contested. Several studieshave examinedwhetherthe bilingual linguistic experience is beneficial to Executive Functions (EF),withdifferentexperimentalparadigmsandinconclusiveresults.Themajorityof those studies tended to focuson cognitive control as a singlemechanismrather than as a multitude of processes dynamically related to each other(Braver 2012; Green and Abutalebi, 2013), and used tasks that yieldeddivergent measures of EF (Paap and Sawi, 2014). Moreover, small samplesizes and non-conservative analytical methods put into question thegeneralizabilityofsuchresults.

32 17

thetransitiveconstruction(e.g.feedthebaby)andtotheindirectobjectinthetransitiveresultativeconstruction(e.g.feedmilktothebaby).

Preliminary results of the analysis suggest that the differences inthematic role distribution affect the ability of the verb feed to generatemetaphorical meaning. In the transitive construction, the verb feed isinterpreted literally each time it is combined with a human PATIENT. ThesourcedomainofNUTRITIONisevokedbytheverbitselfwhileneitherofitsarguments evokes a different conceptual domain. The THEME is leftunexpressedinthetransitiveconstructionbutisassumedtobefood,possiblyincombinationwithwaterorothersubstancesnecessary forahumanbeingtothrive.

Inthetransitiveresultativeconstruction,however,thedirectobjectposition is occupied by the THEME, which is thus explicitly expressed asopposedtojustbeingassumed.Here,theTHEMEcanbeexpressedbyanounevokingaconceptualdomainotherthanNUTRITION,suchasINFORMATIONTRANSFER in They are feeding lies to the people. In such cases, the entireconstruction, including the verb feed is interpreted metaphorically, asreferringtothetransferofinformationratherthansustenance.Theseresultsseem to indicate that the ability of the verb feed to interact with multipleargument-structureconstructionsystematicallycontributestoitspolysemouspotential by allowing it to specify different thematic roles, some of whichfacilitatemetaphoricalextensionintodifferentconceptualdomains.SESSION8 TUESDAY30THMAY15.55-17.1015.55-16.20AscientificrealiststancetowardssociolinguisticmetatheoryJohannesWoschitz•UniversityofEdinburgh

Allegedly, sociolinguistics as a scientific discipline appearedon theacademicsceneinthe1960s.Focusingontherelationshipbetweenlanguageand society, it sought to leave old linguistic paradigms behindwhile at thesame time establishing its ground alongside the Chomskyan school ofgrammar. To do so, scholars from different backgrounds worked togetherinterdisciplinarily,andoutofthiscollaborationbranchescameintobeingthatnow show signs of incipient separatism, leaving sub-disciplines mainlyconcerned with their own study areas that are still commonly subsumedunderthesametermofsociolinguistics.

Withinthescientificcommunity,thisseparatismhasbeenregardedas given, without any awareness of the possible magnitude and impactongoing philosophical tensions below the surface might have on theirmethodology, modes of data interpretation and linguistic theorising as awhole.Intherecent2016bookSociolinguistics:TheoreticalDebateseditedbyCoupland,however,importantscholarsinthefield(Silverstein,Gal,Bell,etc.)have started toaddress this issuebydebunking trivialities in sociolinguisticfindings, arguing that these can be traced back to a shaky commonphilosophical ground. Turning to the plurality of sociolinguistic disciplines,they weigh their suitability for the volume’s newly postulated goal ofsociolinguistics that should ultimately enable it to progress as a unified

31

Inthisstudy,weadministeredtheAX-Continuous PerformanceTask(AX- CPT) to 201 participants to evaluate the existence of a bilingualadvantage.OriginallydevelopedtomeasurecognitivecontrolprocessesinEFdynamically (Braver et al., 2009), this taskwas recently adopted in a studythatcomparedbilingualandmonolingualperformance(Moralesetal.,2013).Here, we tested Italian-Sardinian bilinguals (46), Italian-English bilinguals(54), Italian-SardinianPassivebilinguals (43)and Italianmonolinguals (58).Whenwecomparedaccuracyasindividuallyaggregatedproportions(i.e.eachparticipanthasasinglescoreforeachtypeoftrial),acrossgroups(controllingfor differences in age and years of education), we replicated the results ofMorales et al. (2013): therewas a significant interaction between conditionand group, with Italian-Sardinian participants performing better than theItalianmonolinguals in theAY condition,whichhasbeen claimed to engageinhibitory control (p = .012). The Behavioural Shift Index, a compoundedscore originally proposed by Braver et al. (2009), also showed betterperformanceinthebilingualparticipantsthaninthemonolinguals(p=.032). However, when we analysed the accuracy data using binomialmixed-model regression, which allows for a more accurate random effectsspecification, we found no evidence for a bilingual advantage: pairwisecomparisonbetweengroupsandconditionsdidnotreachsignificance.Theseresults show how studies investigating the bilingual advantage in EF arelimited in their generalizability. Critically, our study demonstrates theimportanceof factoring in individual variabilitywhenstudyingbilingualism,and it highlights three fundamental problems in current research on therelationbetween languageandattention–namely,a theory-drivenselectionof experimental tasks, the choice of analytical strategies as well as limitedsamplesizes.10.50-11.15Effects of conceptual accessibility and similarity in simple sentenceproductioninMandarinYangziZhou,HollyBranigan,&MartinPickering•UniversityofEdinburgh Mandarinhasthreewaystoproduceasimplesentence:

a)Canonicalactive:(Helen)da-le(Johnny)‘(Helen)hit(Johhny)’b)BA-active:(Helen)BAJohnnyda-le‘(Helen)hitJohnny’c)BEI-passive:(Johhny)BEI(Helen)de-le‘JohnnywashitbyHelen’Bock &Warren (1985) proposed that language production can be

affectedbyentities’conceptualaccessibility,referringtotheeasetoretrieveanentity’s concept from long-term memory. Prat-Sala & Branigan (2000)categorized conceptual accessibility into inherent (intrinsic semanticcharacteristics such as animacy) and derived accessibility (temporaryaccessibility gained by contextualmanipulation such as saliency). Effects ofanimacy received convergent results (e.g. McDonald, Bock & Kelly 1993),however,conflictingevidencewasfoundregardingtheeffectsofsaliency(e.g.Osgood1971;Bock&Irwin1980).Inaddition,conceptualsimilaritybetweentheelementscouldalsoconstrainlanguageproduction(e.g.Gennari,Mirkovic&MacDonald2012).

18

Limited studies examined the accessibility and/or the similarityeffects in Mandarin. As a pro- drop language, in Mandarin, word order,grammatical function order and thematic role order can all be varied.Therefore, it is interesting to examine the determinants of languageproduction inMandarin, given all these variations.Hsiao, Gao&MacDonald(2014) reported a similarity effect inMandarin simple sentence productionsuch that more subject omission occurred when the entities weresemanticallysimilarthanwhentheyweredissimilar.However,thesefindingswerequestionablebecausefirst,thematerialsdesignwasbiased(inconsistentintroductory pictures before the target picture). Second, only patient’sanimacy, but not agent’s animacy was manipulated. Thus, whether subjectomissionisduetosimilarityortheanimatepropertyoftheomittedelementscannot be distinguished. Also, some confounding factors, e.g. entities’similarity,werenotcontrolledwell.

Our on-going study investigates the effects of conceptualaccessibility (saliency and animacy) and similarity inMandarin.Hsiao et al.,(2014)andPrat-Sala&Branigan (2000)wereadaptedby includinga largerset of carefully controlled materials. Agents’ and patients’ animacy weremanipulated, and patient was made salient. A picture-description task wasused. It is predicted that if there is a saliency effect,more passive and BA-structuresentencesthanthecanonicalones.Ifthereisananimacyeffect,thenspeakerswilltendtogivesyntacticprominencetoanimateentities.Differentpatterns will also be shown regarding whether subject omission is due tosimilaritybetween theentitiesoranimacy.Basedon theresultsof thepilot,discourse saliency exerted a relatively weaker effect than animacy inMandarin.ThisisincompatiblewithPrat-SalaandBranigan(2000)butagreeswith Sridhar (1988). Similarity effect needs further investigation due tolimitedinstancesofargumentomissions.Afull-scalestudywillbeconductedinAprilandwillbeabletopresentattheconference.11.15-11.40SyntacticprimingandanimacyeffectsinRussian-speakingchildrenAlinaKonradt•UniversityCollegeLondon

Syntactic Priming is a phenomenon characterized by speakers’tendency toproducerecentlyexperiencedsentencestructures.Forexample,uponhearing a passive voice construction (e.g. A catwas chased by a dog),speakers themselves are likely to produce a passive rather than an activesentence(e.g.Adogchasedacat).Alargebodyofevidencesupportstheclaimthat exposure to specific syntactic constructions (primes) subsequentlyfacilitates parsing or production of similar or identical syntactic structures(targets)(Pickering&Ferreira,2008).

Recent experimental work within the syntactic priming paradigmsuggests that the effects of priming are not purely syntactic, but are alsoobservable on the level of information structure, and that both, adults andchildren, are susceptible to these effects (Fleischer, Pickering, & McLean,2012; Vasilyeva&Waterfall, 2012). For example, it is claimed that passivesemphasize the patient of the action, and that hearing a passive sentencewouldpromoteproductionofnon-passiveconstructionswhichhavethesame

(π1),causingher fathertorebukeher forspendinghermoneysocarelessly(π2).-Result(π1,π2)

(3) #Mary enjoyed an extremely expensive meal at the Balmoral

(π1), eating a decadent salmon fillet with a lemon and herbsauce (π2) and causingher father to rebukeher for spendinghermoneysocarelessly(π3).- ???Elaboration(π1,π2)&Elaboration (π1,π3)&Result (π2,π3)

(4) Johnbehaved very badly at the birthdayparty, breaking all oftheglassesonthetable.-breaker=John

(5) John behaved very badly at the birthday party, causing hismotheragreatdealofembarrassment.-breaker=John’sbehavingverybadlyatthebirthdayparty

(6) Johnbehaved very badly at the birthdayparty, breaking all oftheglassesonthetableandcausinghismotheragreatdealofembarrassment.-breaker=John&causer=John

15.20-15.45Construction grammar and lexical polysemy: A case study of the verbfeedintwoargumentstructureconstructionsIvanaKráliková•MasarykUniversity

Muchofworkonlexicalpolysemy(Taylor,2012;Cruse,1995;Gibbs&Matlock,1997)hasstressedthattheultimategenerationofmeaninggreatlydepends on features of linguistic context, including collocation and theconstructions words occur in. Exploring the emergence of metaphoricalmeaning1 in different argument structure constructions, Sullivan (2013)argues thatmost verbs arenot fully semantically specified: they relyon thearguments they take for the ultimate generation of meaning. She furtherpoints out that when an argument structure construction is interpreted asmetaphoric, the target domain is evoked by one or more of the verb‘sargumentswhile the verb itself evokes the source domain, thus lacking theabilitytopromptmetaphoricinterpretationonitsown.

FollowingSullivan’s(2013)insights,thispaperexaminestheEnglishverb feed in two different argument structure construction (in sense ofGoldberg, 1995): the transitive construction (e.g. feed the baby) and thetransitiveresultativeconstruction(e.g.feedmilktothebaby),analysingtheirroleintheemergenceofmetaphoricalmeaning.Basedonacorpusanalysisoftheverbinthetwoconstructions,itfocusesoncaseswherethethematicroleof PATIENT is assumed by a word denoting a human entity, such as child,baby,family,peopleorpopulation.ThePATIENTroleisattributeddifferentlywithinthetwoconstructions:itcorrespondstothedirectobjectoffeedwithin

30 19

discoursefunctionofemphasizingthepatient.Atthesametime,Titov(2017)proposedthatthelinearorderofargumentsinasentenceislicensedbytheirrelativeinterpretiveprominence;andthatanimacyisoneofthefeaturesthatdetermines such prominence. To illustrate, all else being equal, speakerswouldbemorelikelytousestructureswhereananimateentityprecedesaninanimate than vice versa (e.g. Themousewas splashed by the hose ratherthanThehosesplashedthemouse).Itisthereforepossiblethatatleastsomeof the effects that have been interpreted as information structure primingwereinfactdrivenbyunevendistributionofanimacyintargets.

The current experiment had two aims. First, to evaluate theinformation structure priming hypothesis, namely the proposal that inscrambling languages, hearingpassiveprimeswould facilitateproductionofstructures that are syntactically unrelated to the passive, but which have asimilardiscoursefunction(Fleischeret.al.,2012;Vasilyeva&Waterfall,2012).Second, to investigatethe impactofanimacydistribution intargeteventsonthechoiceofstructureselectedbyspeakers.Inordertodrawadirectparallelwith Vasilyeva andWaterfall’s study, Russianspeaking children aged 4 to 7weretested.Therewerethreeconditions:passive,wherepriortodescribingtarget events the participants heard passive primes; active, where activesentenceswereprimed;andbaseline,wherenoprimeswerepresented.Theanimacy in the target events was manipulated: the arguments were eitherequally prominent (both inanimate) or had an unequal prominence(inanimateinstrument/causeananimatepatient/theme).

The results provided no support for the information structurepriminghypothesis:utterancesproducedbythechildrenwerenotaffectedbythe discourse function of emphasizing the patient which passives wereclaimedtocarry.Aneffectofanimacymanipulationwasfoundinthebaselinecondition, providing direct support for Titov's (2017) proposal that relativeanimacy features of arguments predict the choice of word order. Thediscussion highlights potential implications of these findings for the futureexperimentalworkwithinthesyntacticprimingparadigm.11.40-12.05Parsingwh-questions:Evidence fromL1-Greekadults and implicationsforbilinguallearningKaterinaPantoula•UniversityofEdinburgh

Crosslinguistic research shows that monolingual and bilingualspeakers process ambiguous syntactic dependencies differently. Theirdifferences have been attributed to parsing strategies (Clahsen & Felser,2006).More recently, however, it has been demonstrated that they processsyntactic filler-gapdependencies suchaswh-questionsusingsimilarparsingroutines(Pozzan&Trueswell,2015).Crosslinguisticresearchalsoshowsthatobject wh-questions are more difficult than subject wh-questions tocomprehend in typical first language development (Sauerland et al. 2016;inter alia). To date, only one study has investigated how bilingual childrencomprehendwh-questions.RoeschandChondrogianni(2016)studied5-year-old French-German-speaking bilingual children. They found poorlycomprehended sentence-final case-marked German object wh-questions in

to less precise encoding of the form.Bymanipulating the veridicality of therelationship between form andmeaning, I compared learners’ performanceon a sign-repetition task for BSL signs paired with different Englishtranslations. I used the 3D skeletal-tracking capabilities of the MicrosoftKinecttoquantifythetrajectoriesof learners’wristsduringsignproduction.Countertotheresultsoftheaforementionedstudies,Ididnotfindsupportforthe idea that iconic forms are encoded less precisely. I will discuss somepossible reasons for this divergence in results, and possible approaches forfuturework.

SESSION7 TUESDAY30THMAY,14.55-15.4514.55-15.20FreeadjunctsanddiscoursestructureJamesE.M.Reid•UniversityofEdinburgh

It isanemergingconsensuswithinthefieldofdiscoursecoherencethatcoherencerelationscanholdnot justbetweenthecontentofsentences,but also between the content of their constituent parts (e.g. Rohde et al.2011). This talk is concerned with gerund-participial free adjuncts,constructionswhichnotonly sharea coherence relationwith thecontentoftheir matrix clause, but are also syntactically subordinate to their matrixclause.Specifically,IwillexaminetheinteractionbetweendiscoursestructureandsyntacticstructureinexamplesofthisconstructionthatexpressaResultcoherence relation bymeans of a causative verb that takes the content of aclause as its first argument (e.g.Mary lost her keys, forcing her to spend thenight intheshed~Mary’s losingherkeys forcedhertospendthenight intheshed). Accounts of discourse structure (e.g. Asher and Vieu 2005, Vieu andPrévot 2008) have singled out Result as a particularly unstable coherencerelation in termsofwhether it is coordinatingor subordinating, such that itcan be coerced into behaving in a manner suggestive of a subordinatingcoherencerelationiftherearesufficientlystrongcuespresentinthecontext.Iprovide further evidence in support of this view, arguing that discoursesegmentsintroducedbyfreeadjunctsarealwaysdiscourse-subordinatetothesegmentintroducedbytheirmatrixclause.This,incombinationwithexistingtheoretical machinery within Segmented Discourse Representation Theory(Asher1993)andobservationsmadeinrelationtocoordination,allowsforanexplanationtobeputforwardastowhy(i)contentintroducedbycoordinatedfreeadjunctsmustbe interpretedasholdingat thesame ‘level’ofdiscourse(see(1-3)),and(ii)suchconstructionsmustshareacommonantecedent(see(4-6):

(1) Mary enjoyed an extremely expensive meal at the Balmoral(π1), eating a decadent salmon fillet with a lemon and herbsauce(π2).-Elaboration(π1,π2)

(2) Mary enjoyed an extremely expensive meal at the Balmoral

29 20

bilingualchildren.Thesestudieshaveshownthatcertainmorhposyntacticcues can facilitate comprehension. Comprehension of wh-questions hasonly been studied inmonolingual and bilingual first language acquisitionusingoffline techniques.Offlineresearchprovidesuswithpatterns;yet, itdoesnotshowtheprocessesthatunderlieincrementalinterpretation.

In this project, we address the nature of parsing strategiesinvolved in monolingual and bilingual child processing. We investigatebilingual(simultaneousandsequentialL1Greek-L2English)childrenwithL1-Greek as their community language residing in the UK and L1-Greekchildren growing up in Greece comparing the community with the L1-language. The aim is to investigate how bilingual L1 Greek-L2 EnglishchildrenprocessGreekwhich-questionsinthepresenceofovertcase-andgender-markingbyconductingareal-timeexperiment.

Inthispresentation,wereportfindingsfromapilotstudywithtenL1-Greek monolingual adults. Participants were tested on a visual-worldeye-trackingtaskexaminingGreekwhich-questionswheredistributionandnumber of cues was manipulated, adapted from Chondrogianni andSchwartz (2014). Experimental conditions were: (1) a double-cuecondition,wherenounsinbothsentence-initialandsentence-finalpositionwere case-marked (e.g. object: Pjon.MASC.ACC skiuro.MASC.ACC skepazio.MASC.NOM lagos.MASC.NOM. to vradi? “Which squirrel does the rabbitcover in the evening?”), (2) a wh-cue condition, where NP1 was case-marked (e.g. object: Pjon kikno.MASC.ACC filai to elafi.NEUT dipla stopotami?“Whichswandoesthedeerkissnexttotheriver?”),(3)anNP-cuecondition,whereNP2wascase-marked(e.g.object:Pjoliontari.NEUTplenioipopotamos.MASC.NOMmesastopotami?“Whichliondoesthehippowashintheriver?”).

Results showed that the distribution and the number of the cuecontingencies guided the interpretation of subject-extracted and object-extracted Greek referential which-questions. The L1-Greek adults hadceiling performance on the comprehension of which-questions. However,theirRTsrevealedaneffectofcueandimportantly,theirlookingpatternsatendency towards interpreting the first NP in cue-final which-questionswith neuter nouns in sentence initial position as the patient. That is, insubject-extracted sentences like Pjo liontari.NEUT pleni tonipopotamo.MASC.NOMmesa sto potami? (“Which lionwashes the hippo inthe river?”) when the disambiguating case-marking and gender-markingcuearrivedbelatedlyinthesentencetheinterpretationofGreekreferentialsubject which-questions was compromised. We discuss these findings inrelation to their significance for bilingual acquisition in the context ofcurrentsentenceprocessingaccounts.

POSTERSESSION TUESDAY30THMAY12.05-13.05The influence of temporal context on the production of temporalmorphologyinL2speakersofEnglishQingyuanGardner,VickyChondrogianni,&HollyBranigan •University ofEdinburgh

patterns:Antonymsstandinarelationofcontrariness,whiletheirnegationsare related by contradictoriness. Moreover, gradable adjectives can bemappedontoacontinuousversionofthesquareofoppositionspresentedinHorn (2007). These deep analogies with quantifiers indicate that theadjectives in a system are connected by simple semantic operations,correspondingtothequantifiers’externalandinternalnegations.Ishowthatthesemanticoperationofcomplementationforadjectivesplaystherolethatexternal negation has for quantifiers. The standard semantics for gradableadjectives(Kennedy&McNelly2005)doesnotallowforadirectapplicationof internalnegation togradableadjectives, so Idefinea semanticoperation,called“rotation”,whichcreatesantonymsandproducesatthelogicallevelthepatterns typical of contrariness. I sketch some constraints on the rotationoperation.

Finally, I turn to lexicalization patterns. Horn (2007) proposes ageneral rule concerning the lexicalization of the edges on the square ofoppositions: the contradictory of the fundamental form is never lexicalized,while every other form can be lexicalized.While gradable adjectives do notcontradict this generalization, they seem to showa stricterpattern, since insystemsofgradableadjectives thesubalternof the fundamental formisalsoneverlexicalized.IpresentanobjectiontothisobservationandIansweritbyusingthepreviouslydefinedrotationoperation.

14.20-14.45Iseewhatyoudidthere:Theroleoficonicityintheacquisitionofsigns.AshaStewart•UniversityofEdinburgh

Recent theoretical syntheses offer a view of language in whichiconicity–aperceivedresemblancebetweenformandmeaning–isseenasafundamentaldesignfeaturealongsidearbitrariness(Dingemanseet.al.2015).Underthisview,iconicityservestobootstrapacquisition,andthereisalargebody of evidence fromboth spoken and gesturalmodalities confirming thaticonic signs are easier to learn than arbitrary signs (for an overview, seeLockwood & Dingemanse, 2015; Perniss et. al. 2010). However, two recentstudiessuggestamorenuancedpictureoficonicity’scontributiontolearning:InalongitudinalstudyofphonologicaldevelopmentinBritishSignLanguage(BSL) learners, Ortega&Morgan (2015) found that learners produce iconicsignswithlessarticulatoryaccuracythanarbitrarysignsofequalcomplexity.Similarly, in an artificial language learning experiment using a whistledlanguage,Verhoefet.al.(2016)foundwhistleswereproducedlessaccuratelyin a conditionwhere iconicitywas possible compared to a conditionwhereiconicity was disrupted by scrambling the correspondence between signalsandmeanings.Thesetworesultsdrawattentiontothefactthatmoststudieson the effect of iconicity on learninghave focusedon the learnability of themapping between form and meaning, thus potentially obscuring subtletiesrelatingtotheacquisitionoftheform.

InthistalkIwillpresenttheresultsofanexperimentfocusingontheformalpropertiesof learners’productionsof iconicandarbitrarysigns fromBSL,hypothesizingon thebasisof the resultsof theaforementionedstudiesthatwhilsticonicityhelpslearnerstoacquirenewmappings,itmayalsolead

28 21

onsignalproductiongenerallyemergeaccordingtosimplerules.Wesystematicallymanipulated:(a) Initial isolation of subpopulations: we describe 3 possible

isolation levels depending on how fast agents could share the same signsystemoravariantofthesamesystem.

(b)Contentbias:fromnobiastostrongpreferenceforavariant.(c) Coordination biases: from full (egocentric) preference for one's

ownvariantstofull(allocentric)preferenceforothers'variants.Linearmodelsrevealedsignificantmaineffectsandinteractions,e.g.:We found differences between themeans of net signal production

from one level of isolation to another.We also show that, in high isolationpopulations, entropy decreases more slowly. Once the agents potentiallysharethesamerecord,thedifferencesgeneratedbytheisolationblockstendtodecrease.

Content bias slightly amplifies the differences between isolationpopulationtypes.Ontheotherhand,medium-lowvaluesofcoordinationbiasincreasetheeffectsofthepopulationstructures.

Thismodelprovidesasophisticatedunderstandingofhowcomplexinteractions between cognitive biases and population dynamics shape theevolutionofcommunicativevariantsastheyspreadinapopulation.13.55-14.20Squaringaline:Howsimilararegradableadjectivestoquantifiers?FaustoCarcassi•UniversityofEdinburgh

In this talk, I will show that there are deep similarities betweenquantificational expressions and gradable adjectives. Aristotle’s square ofoppositions is a graphical tool, central in medieval logics, to represententailment relationsbetweengroupsof terms. It consistsof a squarewherefour terms have beenmapped onto the four edges. Each term in Aristotle’ssquare is connected to any other term by one of four relations (called“contradictoriness”, “contrariness”, “subcontrariness” and “subalterance”).These four relations simply name certain entailment patterns, rather thanexplainingthem.

Groupsofquantifiers canbemappedon the squareofoppositions,and it is easy to show that they validate the relevant entailment patterns.Theseapparentlydisparateconnectionscanbefurthershowntofollowfromthelogicalformofquantifiers.Whenquantifiersaredefinedasapplicationsofthe semantic operations of external and internal negation on the universalquantifier, the relevant entailment relations can be derived. For example,“none” and “some” instantiate all the entailment patterns characteristic ofcontradictorinessbecause“some”istheexternalnegationof“none”.

Whilequantifiershavetraditionallybeenmappedontothesquareofoppositions,therelationbetweengradableadjectivesandthesquareismuchless understood. Gradable adjectives have two forms, ameasure form – e.g.“Joe is60yearsold”–andapositive form–“Joe isold”.Moreover,gradableadjectivescomeinsystemsofantonyms,e.g.“tall”/“short”,“old”/“young”etc.I show a way of mapping systems of gradable adjectives in their positiveforms onto the square of oppositions that validates the relevant entailment

Adult second language (L2) speakers exhibit variability in L2language production, particularly inflectional morphology (Lardiere, 1998),especially when the L1 does not use inflectional marking. Alternative accountsofoptionalinflectionalmarkingproposedifferentsourcesoffailure,including absence of a representation for tense, or difficulty in processinginflectionalmorphology(Hawkins&Chan,1997;Prevost&White,2000).Theuse of tense appropriate inflections in production has been assumed as anindicator of representational distinctions. The present study investigateswhether temporal context on affects morphological accuracy during L2-English production by L1-Mandarin speakers (whose L1 does not overtlymarktense). 17 Advanced L2 and 17 native English speakers produced scenedescriptions for transitive events. We manipulated both temporal context(Past, Present Habitual) via temporal adverbials (Yesterday, Everyday), andsubject number (singular, plural), to elicit obligatory temporal morphology(past tense –ed (PED), 3SG –s (THS)). Temporal context was a significantpredictor for both PED and THS inflectional markings (p<.001). That is, L2speakersproducedtemporal-context-appropriatemorphemes.Morphologicalaccuracywassignificantlyhigher inthePastthanPresentHabitual temporalcontext (p<.001). The L2 group showed persistent errors in omission andcommission (i.e. incorrect use) of inflections, especially in the PresentHabitualcondition(Figure1). The results demonstrate that L2-English speakers whose L1 doesnotovertlymark tenseare sensitive to temporal cues in theirproductionofinflectionalmorphology. These results therefore provide evidence for thesespeakers’useofatensecategory.Assuchtheyargueagainsttheabsenceofarepresentation for tense (which would wrongly predict absolute omission).However,thehighinflectionalerrorratesindicatethattheseL2speakersusedtemporal cues inconsistently. Moreover, the high number of THS omissionsindicatesaneffectoffeaturalcomposition,wherethefeaturallycomplexTHSinflectionwasespeciallyproblematicinproduction.Theseresultssupportanaccount inwhich theproductionofmorphological inflections ispredicted tobehighlyinfluencedbyitsfeaturalcomplexity.

27 22

Thus, while the languages usually became more informative over

time,itisclearthatthepressurethatshapedthemcameonlyfromlearning.Iwill therefore argue that in these types of experiments it is important toconsider the effects of bothpressures, and that futurework should look fortestcaseswherethesimplicityandinformativenesspressuresarenotaligned.

13.30-13.55Population dynamics effects on the evolution of communicativeconventionsJoséSegoviaMartín&MónicaTamariz•UniversityofEdinburgh

Linguistic conventions in a community spread widely, and lead tothegradualextinctionofotherconventions.Several factorsaffect thespreadofconventions:somerelatetothestructureofthepopulation(e.g.Lupyan&Dale2010),whileothersrelatetocognitivebiasesthataffecttheindividual'slikelihoodofadoptingagivenvariant(content-, frequency-andmodel-basedbiases; Boyd& Richerson 1985). In this studywe investigate the effects onvariant spread of the separate and joint action of population dynamics andcognitivebiases.

We ran computer simulations in eight-agent microsocieties. Atround 0, each agent produces its own unique variant signal; in successiverounds,agents interact inchangingpairs,andeachagentmayproduce theiroriginal variantor anothervariantproducedbyapartner.As somevariantsdisappearwhileothersspread tomultipleagents, theentropyof thevariantset always decreases. Our model shows that population structures have asignificant impact both on the velocity of entropy decline and on the netvariant production of each agent. Furthermore, population dynamics effects

26

IndirectnessinthefaceofculturalcompetencyMayOuma•MeiseiUniversity

Zhao (2013) describes globalization as the “process and theconsequence of shrinking distance”.With the rapid increase ofmigration in recentdecadesandthusthespreadofglobalisation,andthenecessityofinter-culturalcommunication,thereisaneverrisingneedforindividualstopossess“global competency”. What does this actually entail? According to the USNational Education Association, global competency is a way of being, ofpersonhood.Hunteretal.shedsomelightonwhatthis‘wayofbeing’consists,surmisingthatglobalcompetencyisamultitudeofconceptsembodyingskills,knowledge,attitudeanddisposition;withself-awarenessatitscore.

In reading theworksof researchers likePrueHolmesandmyownexperiencestalkingtofriendswhohadlivedabroadortravelledextensively,Idiscovered that they key was not global competency but rather culturalcompetency.InmypaperIexplorethenotionofculturalcompetency;howitis acquired and ten key skills that it births. I also delve into the pitfalls ofsimply teaching language, and also language usage in different societies.Linguistic capability alone does not determine whether an individual canfunctionand/orthriveinaforeign/globalenvironment.

I collectedmy data through variousmethods including interviews,writingfieldnotesandvideorecordings.Thesewerespreadoveraperiodof1year and featured various demographics, businessmen, students, professorsandforeigners living in Japan.Themainareaofmydatathat Iwould liketofocusoninthispresentationareconversationswithmyJapaneseclassmatesandprofessors.IoftenheardthattheJapanese“neversaidwhattheymeant”andIwantedtofindouthowmuchtruthwasinthisstatement.WhatIfoundwasfascinatingandIbelieveprovidesavaluableinsightintothetopicofnotonlyculturalcompetency,butalsoindirectnessstudies.Phonetic preaspiration of word-final voiceless fricatives in North EastScotlandDavidA.J.Warren•UniversityofAberdeen Phonological preaspiration of stops occurs in varieties of ScottishGaelic (Lade- foged, Ladefoged, Turk, Hind, & Skilton, 1998), and phonetic(‘non-normative’(Gordeeva&Scobbie,2010,167))preaspirationofvoicelessstops and fricatives has been attested to inNewcastle (Docherty& Foulkes,1999)andTynesideEn-glishmoregenerally(Jones&Llamas,2003).InScotsand Scottish Standard English (SSE), voiceless stops are frequently realisedwith glottal reinforcement (Scobbie, Gordeeva, & Matthews, 2006; Wells,1982) and, indeed, ‘there have been no reports of wide and early glottalabduction beforeword-final stops in SSE that could result in preaspiration’(Gordeeva & Scobbie, 2010, 169). Phonetic preaspiration before word-finalvoiceless fricatives, however, has been observed. Gordeeva and Scobbie(2010)reportfindingsforspeakersofSSE,allbutoneofwhomwasnativetoScotland’s Central Belt. Voiceless fricatives were preaspi- rated 48% of thetime,withwomenusingpreaspiratedvariantsmorefrequently,althoughthis

23

gradeandgender; (d)genderandgradehavean impactonFLachievement,with girls achieving higher scores than boys and lower graders achievinghigher scores; (e) a stronger negative relationship between FL anxiety andformal assessment are found compared with its relation with regular unitassessment;(f)FLanxietyresultinpoorFLperformanceandtheinfluenceofFLanxietybecomesstrongeras their instructional level increase; (g)of fourcomponents of FL anxiety, communication apprehension appears to be thestrongestpredictorofFLachievement.

SESSION6 TUESDAY30THMAY,13.05-14.4513.05-13.30IteratedlearningoptimizesforsimplicityJonW.Carr•UniversityofEdinburgh

Language is shaped by the pressures of simplicity andinformativeness:Alanguagemustbebothsimple(learnable)andinformative(communicatively useful) if it is to be propagated (Kemp & Regier, 2012;Kirby,Tamariz,Cornish,&Smith,2015).However, itcanbedifficulttoteasethese two pressures apart because the features that make a languagelearnableareoften thesame features thatmakea language informative.Forexample, Carstensen, Xu, Smith, and Regier (2015) conducted an iteratedlearning experiment in which participants learned labels for spatialrelationships, and they found thatover generations the languages tended tobecome more informative. But since this experiment included nocommunicative pressure, another explanation can be provided in terms ofsimplicity.InthistalkIwillpresentthreeexperimentsthataimtoshedsomelightonthis.

In Experiment 1, participants were asked to learn and recall apartitionofa two-dimensionalmeaningspace into foursemanticcategories;in the first condition, thepartitiononlymarked the angledimension; in thesecond, the partition onlymarked the size dimension; and in the third, thepartition marked both dimensions simultaneously (see top of Fig. 1). Wefound that partitions marking both dimensions were harder to learn thanpartitions marking only one dimension. One dimensional category systemsarefavouredbylearning.

In Experiment 2, the test procedure was modified: Rather thanproducelabelsforstimuli,participantswereaskedtoselectstimuliforlabels.Together,thesefirsttwoexperimentsprovidedataabouttheproductionandcomprehensionfacetsofcommunication,allowingustosimulatewhatwouldhappeninacommunicativescenario.Wefoundthatcommunicativeaccuracywas highest when participants used the partition that marked bothdimensions. Two dimensional category systems are favoured bycommunication.

InExperiment3,participantshadto learnthepartition inferredbythepreviousparticipant inan iterated learning chain (seebottomofFig.1).All chains fixated on highly simplified partitions that never marked bothdimensionssimultaneously.

differencewasnotstatisticallysignificant.Precedinghighvowels/iu/wereassociated with lower rates of fricative preaspiration than lower vowels,with fricatives after /a/ being preaspirated most consistently. Phrasalposition was also important, with tokens in phrase-final position beingpreaspiratedmoreoftenthanthoseinnon-finalposition.

ThereisnoexistingliteratureonpreaspirationinScotlandoutsidetheCentralBelt,andthestatusofthefeatureforspeakersofanyvarietyofScots isnotknown.Oneof the subjects inGordeevaandScobbie’s (2010)studywasborninAberdeenandusedpreaspiratedvariantsapproximately85% of the time, the second-highest of their subjects.While this may beideolectal, it raises thepossibility thatprea- spiratedvariantsoccurmorefrequently in other regions of Scotland. This paper presents findingsregarding the presence of phonetic preaspiration amongst speak- ers ofScotsandScottishEnglish fromtheNorthEastofScotland,comparing thefeature in both continuous speech and single-word lists. Preaspiratedvariants occur more frequently in word-list and phrase-final position incontinuousspeechthanintheCentralBelt,butthevariantisdispreferredinnon-finalposition.AsperGordeevaandScobbie(2010),fricativesfollowingthehigh vowels are somewhat resistant topreaspiration; however, in thesingle-word list themajorityof thesetokenswerestillpreaspirated. Ialsobriefly present findings for the presence of consistent phoneticpreaspirationbeforevoicelessstopsforseveralspeakers.Exploringyounglearners'foreignlanguageanxietyinChinaXiaoyiHu•UniversityofEdinburgh FLanxietyhaslongbeenregardedasamajorobstacleinlanguageacquisition and many of the previous related research contends that FLanxiety exist among adult learners. However, to date, whether the samelevelsandpatternsofFLanxietyarepresentamongchildrenandwhetherthe same relationship with FL achievement is found is unknown. Inaddition, research studying the relation between FL anxiety and FLachievement does not differentiate between achievements tested underexamconditionsvsmoreregularassessments.Also,thereislittleresearchexploring the interactive effect of FL anxiety, achievement and studentscharacteristics(gradeandgender)amongchildren.

In this study, 631 pupils aged 9 to 12 from grade 4-6 in Chinaparticipated. Participants’ FL anxiety was assessed on an adapted child-friendly version of FLCAS (foreign language classroom anxiety scale)proposed by Horwitz, Horwitz and Cope (1986) and their FL proficiencywas assessed through their standardized formal tests and regular unitpaper exams. The role of FL anxiety as well as its four components isexaminedinrelationtochildren’sFLachievement,gender,andgrade.

Thefindingrevealedthat(a)communicationapprehension,fearofnegativeevaluation,testanxietyandattitudetowardsclassroomconsistsofFLanxiety;(b)thelevelofFLanxietyincreasesasgradeincreases,butdoesnot change with the effect of gender; (c) of the four components of FLanxiety, only communication apprehension and test anxiety differ across

24 25

gradeandgender; (d)genderandgradehavean impactonFLachievement,with girls achieving higher scores than boys and lower graders achievinghigher scores; (e) a stronger negative relationship between FL anxiety andformal assessment are found compared with its relation with regular unitassessment;(f)FLanxietyresultinpoorFLperformanceandtheinfluenceofFLanxietybecomesstrongeras their instructional level increase; (g)of fourcomponents of FL anxiety, communication apprehension appears to be thestrongestpredictorofFLachievement.

SESSION6 TUESDAY30THMAY,13.05-14.4513.05-13.30IteratedlearningoptimizesforsimplicityJonW.Carr•UniversityofEdinburgh

Language is shaped by the pressures of simplicity andinformativeness:Alanguagemustbebothsimple(learnable)andinformative(communicatively useful) if it is to be propagated (Kemp & Regier, 2012;Kirby,Tamariz,Cornish,&Smith,2015).However, itcanbedifficulttoteasethese two pressures apart because the features that make a languagelearnableareoften thesame features thatmakea language informative.Forexample, Carstensen, Xu, Smith, and Regier (2015) conducted an iteratedlearning experiment in which participants learned labels for spatialrelationships, and they found thatover generations the languages tended tobecome more informative. But since this experiment included nocommunicative pressure, another explanation can be provided in terms ofsimplicity.InthistalkIwillpresentthreeexperimentsthataimtoshedsomelightonthis.

In Experiment 1, participants were asked to learn and recall apartitionofa two-dimensionalmeaningspace into foursemanticcategories;in the first condition, thepartitiononlymarked the angledimension; in thesecond, the partition onlymarked the size dimension; and in the third, thepartition marked both dimensions simultaneously (see top of Fig. 1). Wefound that partitions marking both dimensions were harder to learn thanpartitions marking only one dimension. One dimensional category systemsarefavouredbylearning.

In Experiment 2, the test procedure was modified: Rather thanproducelabelsforstimuli,participantswereaskedtoselectstimuliforlabels.Together,thesefirsttwoexperimentsprovidedataabouttheproductionandcomprehensionfacetsofcommunication,allowingustosimulatewhatwouldhappeninacommunicativescenario.Wefoundthatcommunicativeaccuracywas highest when participants used the partition that marked bothdimensions. Two dimensional category systems are favoured bycommunication.

InExperiment3,participantshadto learnthepartition inferredbythepreviousparticipant inan iterated learning chain (seebottomofFig.1).All chains fixated on highly simplified partitions that never marked bothdimensionssimultaneously.

differencewasnotstatisticallysignificant.Precedinghighvowels/iu/wereassociated with lower rates of fricative preaspiration than lower vowels,with fricatives after /a/ being preaspirated most consistently. Phrasalposition was also important, with tokens in phrase-final position beingpreaspiratedmoreoftenthanthoseinnon-finalposition.

ThereisnoexistingliteratureonpreaspirationinScotlandoutsidetheCentralBelt,andthestatusofthefeatureforspeakersofanyvarietyofScots isnotknown.Oneof the subjects inGordeevaandScobbie’s (2010)studywasborninAberdeenandusedpreaspiratedvariantsapproximately85% of the time, the second-highest of their subjects.While this may beideolectal, it raises thepossibility thatprea- spiratedvariantsoccurmorefrequently in other regions of Scotland. This paper presents findingsregarding the presence of phonetic preaspiration amongst speak- ers ofScotsandScottishEnglish fromtheNorthEastofScotland,comparing thefeature in both continuous speech and single-word lists. Preaspiratedvariants occur more frequently in word-list and phrase-final position incontinuousspeechthanintheCentralBelt,butthevariantisdispreferredinnon-finalposition.AsperGordeevaandScobbie(2010),fricativesfollowingthehigh vowels are somewhat resistant topreaspiration; however, in thesingle-word list themajorityof thesetokenswerestillpreaspirated. Ialsobriefly present findings for the presence of consistent phoneticpreaspirationbeforevoicelessstopsforseveralspeakers.Exploringyounglearners'foreignlanguageanxietyinChinaXiaoyiHu•UniversityofEdinburgh FLanxietyhaslongbeenregardedasamajorobstacleinlanguageacquisition and many of the previous related research contends that FLanxiety exist among adult learners. However, to date, whether the samelevelsandpatternsofFLanxietyarepresentamongchildrenandwhetherthe same relationship with FL achievement is found is unknown. Inaddition, research studying the relation between FL anxiety and FLachievement does not differentiate between achievements tested underexamconditionsvsmoreregularassessments.Also,thereislittleresearchexploring the interactive effect of FL anxiety, achievement and studentscharacteristics(gradeandgender)amongchildren.

In this study, 631 pupils aged 9 to 12 from grade 4-6 in Chinaparticipated. Participants’ FL anxiety was assessed on an adapted child-friendly version of FLCAS (foreign language classroom anxiety scale)proposed by Horwitz, Horwitz and Cope (1986) and their FL proficiencywas assessed through their standardized formal tests and regular unitpaper exams. The role of FL anxiety as well as its four components isexaminedinrelationtochildren’sFLachievement,gender,andgrade.

Thefindingrevealedthat(a)communicationapprehension,fearofnegativeevaluation,testanxietyandattitudetowardsclassroomconsistsofFLanxiety;(b)thelevelofFLanxietyincreasesasgradeincreases,butdoesnot change with the effect of gender; (c) of the four components of FLanxiety, only communication apprehension and test anxiety differ across

24 25

Thus, while the languages usually became more informative over

time,itisclearthatthepressurethatshapedthemcameonlyfromlearning.Iwill therefore argue that in these types of experiments it is important toconsider the effects of bothpressures, and that futurework should look fortestcaseswherethesimplicityandinformativenesspressuresarenotaligned.

13.30-13.55Population dynamics effects on the evolution of communicativeconventionsJoséSegoviaMartín&MónicaTamariz•UniversityofEdinburgh

Linguistic conventions in a community spread widely, and lead tothegradualextinctionofotherconventions.Several factorsaffect thespreadofconventions:somerelatetothestructureofthepopulation(e.g.Lupyan&Dale2010),whileothersrelatetocognitivebiasesthataffecttheindividual'slikelihoodofadoptingagivenvariant(content-, frequency-andmodel-basedbiases; Boyd& Richerson 1985). In this studywe investigate the effects onvariant spread of the separate and joint action of population dynamics andcognitivebiases.

We ran computer simulations in eight-agent microsocieties. Atround 0, each agent produces its own unique variant signal; in successiverounds,agents interact inchangingpairs,andeachagentmayproduce theiroriginal variantor anothervariantproducedbyapartner.As somevariantsdisappearwhileothersspread tomultipleagents, theentropyof thevariantset always decreases. Our model shows that population structures have asignificant impact both on the velocity of entropy decline and on the netvariant production of each agent. Furthermore, population dynamics effects

26

IndirectnessinthefaceofculturalcompetencyMayOuma•MeiseiUniversity

Zhao (2013) describes globalization as the “process and theconsequence of shrinking distance”.With the rapid increase ofmigration in recentdecadesandthusthespreadofglobalisation,andthenecessityofinter-culturalcommunication,thereisaneverrisingneedforindividualstopossess“global competency”. What does this actually entail? According to the USNational Education Association, global competency is a way of being, ofpersonhood.Hunteretal.shedsomelightonwhatthis‘wayofbeing’consists,surmisingthatglobalcompetencyisamultitudeofconceptsembodyingskills,knowledge,attitudeanddisposition;withself-awarenessatitscore.

In reading theworksof researchers likePrueHolmesandmyownexperiencestalkingtofriendswhohadlivedabroadortravelledextensively,Idiscovered that they key was not global competency but rather culturalcompetency.InmypaperIexplorethenotionofculturalcompetency;howitis acquired and ten key skills that it births. I also delve into the pitfalls ofsimply teaching language, and also language usage in different societies.Linguistic capability alone does not determine whether an individual canfunctionand/orthriveinaforeign/globalenvironment.

I collectedmy data through variousmethods including interviews,writingfieldnotesandvideorecordings.Thesewerespreadoveraperiodof1year and featured various demographics, businessmen, students, professorsandforeigners living in Japan.Themainareaofmydatathat Iwould liketofocusoninthispresentationareconversationswithmyJapaneseclassmatesandprofessors.IoftenheardthattheJapanese“neversaidwhattheymeant”andIwantedtofindouthowmuchtruthwasinthisstatement.WhatIfoundwasfascinatingandIbelieveprovidesavaluableinsightintothetopicofnotonlyculturalcompetency,butalsoindirectnessstudies.Phonetic preaspiration of word-final voiceless fricatives in North EastScotlandDavidA.J.Warren•UniversityofAberdeen Phonological preaspiration of stops occurs in varieties of ScottishGaelic (Lade- foged, Ladefoged, Turk, Hind, & Skilton, 1998), and phonetic(‘non-normative’(Gordeeva&Scobbie,2010,167))preaspirationofvoicelessstops and fricatives has been attested to inNewcastle (Docherty& Foulkes,1999)andTynesideEn-glishmoregenerally(Jones&Llamas,2003).InScotsand Scottish Standard English (SSE), voiceless stops are frequently realisedwith glottal reinforcement (Scobbie, Gordeeva, & Matthews, 2006; Wells,1982) and, indeed, ‘there have been no reports of wide and early glottalabduction beforeword-final stops in SSE that could result in preaspiration’(Gordeeva & Scobbie, 2010, 169). Phonetic preaspiration before word-finalvoiceless fricatives, however, has been observed. Gordeeva and Scobbie(2010)reportfindingsforspeakersofSSE,allbutoneofwhomwasnativetoScotland’s Central Belt. Voiceless fricatives were preaspi- rated 48% of thetime,withwomenusingpreaspiratedvariantsmorefrequently,althoughthis

23

onsignalproductiongenerallyemergeaccordingtosimplerules.Wesystematicallymanipulated:(a) Initial isolation of subpopulations: we describe 3 possible

isolation levels depending on how fast agents could share the same signsystemoravariantofthesamesystem.

(b)Contentbias:fromnobiastostrongpreferenceforavariant.(c) Coordination biases: from full (egocentric) preference for one's

ownvariantstofull(allocentric)preferenceforothers'variants.Linearmodelsrevealedsignificantmaineffectsandinteractions,e.g.:We found differences between themeans of net signal production

from one level of isolation to another.We also show that, in high isolationpopulations, entropy decreases more slowly. Once the agents potentiallysharethesamerecord,thedifferencesgeneratedbytheisolationblockstendtodecrease.

Content bias slightly amplifies the differences between isolationpopulationtypes.Ontheotherhand,medium-lowvaluesofcoordinationbiasincreasetheeffectsofthepopulationstructures.

Thismodelprovidesasophisticatedunderstandingofhowcomplexinteractions between cognitive biases and population dynamics shape theevolutionofcommunicativevariantsastheyspreadinapopulation.13.55-14.20Squaringaline:Howsimilararegradableadjectivestoquantifiers?FaustoCarcassi•UniversityofEdinburgh

In this talk, I will show that there are deep similarities betweenquantificational expressions and gradable adjectives. Aristotle’s square ofoppositions is a graphical tool, central in medieval logics, to represententailment relationsbetweengroupsof terms. It consistsof a squarewherefour terms have beenmapped onto the four edges. Each term in Aristotle’ssquare is connected to any other term by one of four relations (called“contradictoriness”, “contrariness”, “subcontrariness” and “subalterance”).These four relations simply name certain entailment patterns, rather thanexplainingthem.

Groupsofquantifiers canbemappedon the squareofoppositions,and it is easy to show that they validate the relevant entailment patterns.Theseapparentlydisparateconnectionscanbefurthershowntofollowfromthelogicalformofquantifiers.Whenquantifiersaredefinedasapplicationsofthe semantic operations of external and internal negation on the universalquantifier, the relevant entailment relations can be derived. For example,“none” and “some” instantiate all the entailment patterns characteristic ofcontradictorinessbecause“some”istheexternalnegationof“none”.

Whilequantifiershavetraditionallybeenmappedontothesquareofoppositions,therelationbetweengradableadjectivesandthesquareismuchless understood. Gradable adjectives have two forms, ameasure form – e.g.“Joe is60yearsold”–andapositive form–“Joe isold”.Moreover,gradableadjectivescomeinsystemsofantonyms,e.g.“tall”/“short”,“old”/“young”etc.I show a way of mapping systems of gradable adjectives in their positiveforms onto the square of oppositions that validates the relevant entailment

Adult second language (L2) speakers exhibit variability in L2language production, particularly inflectional morphology (Lardiere, 1998),especially when the L1 does not use inflectional marking. Alternative accountsofoptionalinflectionalmarkingproposedifferentsourcesoffailure,including absence of a representation for tense, or difficulty in processinginflectionalmorphology(Hawkins&Chan,1997;Prevost&White,2000).Theuse of tense appropriate inflections in production has been assumed as anindicator of representational distinctions. The present study investigateswhether temporal context on affects morphological accuracy during L2-English production by L1-Mandarin speakers (whose L1 does not overtlymarktense). 17 Advanced L2 and 17 native English speakers produced scenedescriptions for transitive events. We manipulated both temporal context(Past, Present Habitual) via temporal adverbials (Yesterday, Everyday), andsubject number (singular, plural), to elicit obligatory temporal morphology(past tense –ed (PED), 3SG –s (THS)). Temporal context was a significantpredictor for both PED and THS inflectional markings (p<.001). That is, L2speakersproducedtemporal-context-appropriatemorphemes.Morphologicalaccuracywassignificantlyhigher inthePastthanPresentHabitual temporalcontext (p<.001). The L2 group showed persistent errors in omission andcommission (i.e. incorrect use) of inflections, especially in the PresentHabitualcondition(Figure1). The results demonstrate that L2-English speakers whose L1 doesnotovertlymark tenseare sensitive to temporal cues in theirproductionofinflectionalmorphology. These results therefore provide evidence for thesespeakers’useofatensecategory.Assuchtheyargueagainsttheabsenceofarepresentation for tense (which would wrongly predict absolute omission).However,thehighinflectionalerrorratesindicatethattheseL2speakersusedtemporal cues inconsistently. Moreover, the high number of THS omissionsindicatesaneffectoffeaturalcomposition,wherethefeaturallycomplexTHSinflectionwasespeciallyproblematicinproduction.Theseresultssupportanaccount inwhich theproductionofmorphological inflections ispredicted tobehighlyinfluencedbyitsfeaturalcomplexity.

27 22

bilingualchildren.Thesestudieshaveshownthatcertainmorhposyntacticcues can facilitate comprehension. Comprehension of wh-questions hasonly been studied inmonolingual and bilingual first language acquisitionusingoffline techniques.Offlineresearchprovidesuswithpatterns;yet, itdoesnotshowtheprocessesthatunderlieincrementalinterpretation.

In this project, we address the nature of parsing strategiesinvolved in monolingual and bilingual child processing. We investigatebilingual(simultaneousandsequentialL1Greek-L2English)childrenwithL1-Greek as their community language residing in the UK and L1-Greekchildren growing up in Greece comparing the community with the L1-language. The aim is to investigate how bilingual L1 Greek-L2 EnglishchildrenprocessGreekwhich-questionsinthepresenceofovertcase-andgender-markingbyconductingareal-timeexperiment.

Inthispresentation,wereportfindingsfromapilotstudywithtenL1-Greek monolingual adults. Participants were tested on a visual-worldeye-trackingtaskexaminingGreekwhich-questionswheredistributionandnumber of cues was manipulated, adapted from Chondrogianni andSchwartz (2014). Experimental conditions were: (1) a double-cuecondition,wherenounsinbothsentence-initialandsentence-finalpositionwere case-marked (e.g. object: Pjon.MASC.ACC skiuro.MASC.ACC skepazio.MASC.NOM lagos.MASC.NOM. to vradi? “Which squirrel does the rabbitcover in the evening?”), (2) a wh-cue condition, where NP1 was case-marked (e.g. object: Pjon kikno.MASC.ACC filai to elafi.NEUT dipla stopotami?“Whichswandoesthedeerkissnexttotheriver?”),(3)anNP-cuecondition,whereNP2wascase-marked(e.g.object:Pjoliontari.NEUTplenioipopotamos.MASC.NOMmesastopotami?“Whichliondoesthehippowashintheriver?”).

Results showed that the distribution and the number of the cuecontingencies guided the interpretation of subject-extracted and object-extracted Greek referential which-questions. The L1-Greek adults hadceiling performance on the comprehension of which-questions. However,theirRTsrevealedaneffectofcueandimportantly,theirlookingpatternsatendency towards interpreting the first NP in cue-final which-questionswith neuter nouns in sentence initial position as the patient. That is, insubject-extracted sentences like Pjo liontari.NEUT pleni tonipopotamo.MASC.NOMmesa sto potami? (“Which lionwashes the hippo inthe river?”) when the disambiguating case-marking and gender-markingcuearrivedbelatedlyinthesentencetheinterpretationofGreekreferentialsubject which-questions was compromised. We discuss these findings inrelation to their significance for bilingual acquisition in the context ofcurrentsentenceprocessingaccounts.

POSTERSESSION TUESDAY30THMAY12.05-13.05The influence of temporal context on the production of temporalmorphologyinL2speakersofEnglishQingyuanGardner,VickyChondrogianni,&HollyBranigan •University ofEdinburgh

patterns:Antonymsstandinarelationofcontrariness,whiletheirnegationsare related by contradictoriness. Moreover, gradable adjectives can bemappedontoacontinuousversionofthesquareofoppositionspresentedinHorn (2007). These deep analogies with quantifiers indicate that theadjectives in a system are connected by simple semantic operations,correspondingtothequantifiers’externalandinternalnegations.Ishowthatthesemanticoperationofcomplementationforadjectivesplaystherolethatexternal negation has for quantifiers. The standard semantics for gradableadjectives(Kennedy&McNelly2005)doesnotallowforadirectapplicationof internalnegation togradableadjectives, so Idefinea semanticoperation,called“rotation”,whichcreatesantonymsandproducesatthelogicallevelthepatterns typical of contrariness. I sketch some constraints on the rotationoperation.

Finally, I turn to lexicalization patterns. Horn (2007) proposes ageneral rule concerning the lexicalization of the edges on the square ofoppositions: the contradictory of the fundamental form is never lexicalized,while every other form can be lexicalized.While gradable adjectives do notcontradict this generalization, they seem to showa stricterpattern, since insystemsofgradableadjectives thesubalternof the fundamental formisalsoneverlexicalized.IpresentanobjectiontothisobservationandIansweritbyusingthepreviouslydefinedrotationoperation.

14.20-14.45Iseewhatyoudidthere:Theroleoficonicityintheacquisitionofsigns.AshaStewart•UniversityofEdinburgh

Recent theoretical syntheses offer a view of language in whichiconicity–aperceivedresemblancebetweenformandmeaning–isseenasafundamentaldesignfeaturealongsidearbitrariness(Dingemanseet.al.2015).Underthisview,iconicityservestobootstrapacquisition,andthereisalargebody of evidence fromboth spoken and gesturalmodalities confirming thaticonic signs are easier to learn than arbitrary signs (for an overview, seeLockwood & Dingemanse, 2015; Perniss et. al. 2010). However, two recentstudiessuggestamorenuancedpictureoficonicity’scontributiontolearning:InalongitudinalstudyofphonologicaldevelopmentinBritishSignLanguage(BSL) learners, Ortega&Morgan (2015) found that learners produce iconicsignswithlessarticulatoryaccuracythanarbitrarysignsofequalcomplexity.Similarly, in an artificial language learning experiment using a whistledlanguage,Verhoefet.al.(2016)foundwhistleswereproducedlessaccuratelyin a conditionwhere iconicitywas possible compared to a conditionwhereiconicity was disrupted by scrambling the correspondence between signalsandmeanings.Thesetworesultsdrawattentiontothefactthatmoststudieson the effect of iconicity on learninghave focusedon the learnability of themapping between form and meaning, thus potentially obscuring subtletiesrelatingtotheacquisitionoftheform.

InthistalkIwillpresenttheresultsofanexperimentfocusingontheformalpropertiesof learners’productionsof iconicandarbitrarysigns fromBSL,hypothesizingon thebasisof the resultsof theaforementionedstudiesthatwhilsticonicityhelpslearnerstoacquirenewmappings,itmayalsolead

28 21

discoursefunctionofemphasizingthepatient.Atthesametime,Titov(2017)proposedthatthelinearorderofargumentsinasentenceislicensedbytheirrelativeinterpretiveprominence;andthatanimacyisoneofthefeaturesthatdetermines such prominence. To illustrate, all else being equal, speakerswouldbemorelikelytousestructureswhereananimateentityprecedesaninanimate than vice versa (e.g. Themousewas splashed by the hose ratherthanThehosesplashedthemouse).Itisthereforepossiblethatatleastsomeof the effects that have been interpreted as information structure primingwereinfactdrivenbyunevendistributionofanimacyintargets.

The current experiment had two aims. First, to evaluate theinformation structure priming hypothesis, namely the proposal that inscrambling languages, hearingpassiveprimeswould facilitateproductionofstructures that are syntactically unrelated to the passive, but which have asimilardiscoursefunction(Fleischeret.al.,2012;Vasilyeva&Waterfall,2012).Second, to investigatethe impactofanimacydistribution intargeteventsonthechoiceofstructureselectedbyspeakers.Inordertodrawadirectparallelwith Vasilyeva andWaterfall’s study, Russianspeaking children aged 4 to 7weretested.Therewerethreeconditions:passive,wherepriortodescribingtarget events the participants heard passive primes; active, where activesentenceswereprimed;andbaseline,wherenoprimeswerepresented.Theanimacy in the target events was manipulated: the arguments were eitherequally prominent (both inanimate) or had an unequal prominence(inanimateinstrument/causeananimatepatient/theme).

The results provided no support for the information structurepriminghypothesis:utterancesproducedbythechildrenwerenotaffectedbythe discourse function of emphasizing the patient which passives wereclaimedtocarry.Aneffectofanimacymanipulationwasfoundinthebaselinecondition, providing direct support for Titov's (2017) proposal that relativeanimacy features of arguments predict the choice of word order. Thediscussion highlights potential implications of these findings for the futureexperimentalworkwithinthesyntacticprimingparadigm.11.40-12.05Parsingwh-questions:Evidence fromL1-Greekadults and implicationsforbilinguallearningKaterinaPantoula•UniversityofEdinburgh

Crosslinguistic research shows that monolingual and bilingualspeakers process ambiguous syntactic dependencies differently. Theirdifferences have been attributed to parsing strategies (Clahsen & Felser,2006).More recently, however, it has been demonstrated that they processsyntactic filler-gapdependencies suchaswh-questionsusingsimilarparsingroutines(Pozzan&Trueswell,2015).Crosslinguisticresearchalsoshowsthatobject wh-questions are more difficult than subject wh-questions tocomprehend in typical first language development (Sauerland et al. 2016;inter alia). To date, only one study has investigated how bilingual childrencomprehendwh-questions.RoeschandChondrogianni(2016)studied5-year-old French-German-speaking bilingual children. They found poorlycomprehended sentence-final case-marked German object wh-questions in

to less precise encoding of the form.Bymanipulating the veridicality of therelationship between form andmeaning, I compared learners’ performanceon a sign-repetition task for BSL signs paired with different Englishtranslations. I used the 3D skeletal-tracking capabilities of the MicrosoftKinecttoquantifythetrajectoriesof learners’wristsduringsignproduction.Countertotheresultsoftheaforementionedstudies,Ididnotfindsupportforthe idea that iconic forms are encoded less precisely. I will discuss somepossible reasons for this divergence in results, and possible approaches forfuturework.

SESSION7 TUESDAY30THMAY,14.55-15.4514.55-15.20FreeadjunctsanddiscoursestructureJamesE.M.Reid•UniversityofEdinburgh

It isanemergingconsensuswithinthefieldofdiscoursecoherencethatcoherencerelationscanholdnot justbetweenthecontentofsentences,but also between the content of their constituent parts (e.g. Rohde et al.2011). This talk is concerned with gerund-participial free adjuncts,constructionswhichnotonly sharea coherence relationwith thecontentoftheir matrix clause, but are also syntactically subordinate to their matrixclause.Specifically,IwillexaminetheinteractionbetweendiscoursestructureandsyntacticstructureinexamplesofthisconstructionthatexpressaResultcoherence relation bymeans of a causative verb that takes the content of aclause as its first argument (e.g.Mary lost her keys, forcing her to spend thenight intheshed~Mary’s losingherkeys forcedhertospendthenight intheshed). Accounts of discourse structure (e.g. Asher and Vieu 2005, Vieu andPrévot 2008) have singled out Result as a particularly unstable coherencerelation in termsofwhether it is coordinatingor subordinating, such that itcan be coerced into behaving in a manner suggestive of a subordinatingcoherencerelationiftherearesufficientlystrongcuespresentinthecontext.Iprovide further evidence in support of this view, arguing that discoursesegmentsintroducedbyfreeadjunctsarealwaysdiscourse-subordinatetothesegmentintroducedbytheirmatrixclause.This,incombinationwithexistingtheoretical machinery within Segmented Discourse Representation Theory(Asher1993)andobservationsmadeinrelationtocoordination,allowsforanexplanationtobeputforwardastowhy(i)contentintroducedbycoordinatedfreeadjunctsmustbe interpretedasholdingat thesame ‘level’ofdiscourse(see(1-3)),and(ii)suchconstructionsmustshareacommonantecedent(see(4-6):

(1) Mary enjoyed an extremely expensive meal at the Balmoral(π1), eating a decadent salmon fillet with a lemon and herbsauce(π2).-Elaboration(π1,π2)

(2) Mary enjoyed an extremely expensive meal at the Balmoral

29 20

Limited studies examined the accessibility and/or the similarityeffects in Mandarin. As a pro- drop language, in Mandarin, word order,grammatical function order and thematic role order can all be varied.Therefore, it is interesting to examine the determinants of languageproduction inMandarin, given all these variations.Hsiao, Gao&MacDonald(2014) reported a similarity effect inMandarin simple sentence productionsuch that more subject omission occurred when the entities weresemanticallysimilarthanwhentheyweredissimilar.However,thesefindingswerequestionablebecausefirst,thematerialsdesignwasbiased(inconsistentintroductory pictures before the target picture). Second, only patient’sanimacy, but not agent’s animacy was manipulated. Thus, whether subjectomissionisduetosimilarityortheanimatepropertyoftheomittedelementscannot be distinguished. Also, some confounding factors, e.g. entities’similarity,werenotcontrolledwell.

Our on-going study investigates the effects of conceptualaccessibility (saliency and animacy) and similarity inMandarin.Hsiao et al.,(2014)andPrat-Sala&Branigan (2000)wereadaptedby includinga largerset of carefully controlled materials. Agents’ and patients’ animacy weremanipulated, and patient was made salient. A picture-description task wasused. It is predicted that if there is a saliency effect,more passive and BA-structuresentencesthanthecanonicalones.Ifthereisananimacyeffect,thenspeakerswilltendtogivesyntacticprominencetoanimateentities.Differentpatterns will also be shown regarding whether subject omission is due tosimilaritybetween theentitiesoranimacy.Basedon theresultsof thepilot,discourse saliency exerted a relatively weaker effect than animacy inMandarin.ThisisincompatiblewithPrat-SalaandBranigan(2000)butagreeswith Sridhar (1988). Similarity effect needs further investigation due tolimitedinstancesofargumentomissions.Afull-scalestudywillbeconductedinAprilandwillbeabletopresentattheconference.11.15-11.40SyntacticprimingandanimacyeffectsinRussian-speakingchildrenAlinaKonradt•UniversityCollegeLondon

Syntactic Priming is a phenomenon characterized by speakers’tendency toproducerecentlyexperiencedsentencestructures.Forexample,uponhearing a passive voice construction (e.g. A catwas chased by a dog),speakers themselves are likely to produce a passive rather than an activesentence(e.g.Adogchasedacat).Alargebodyofevidencesupportstheclaimthat exposure to specific syntactic constructions (primes) subsequentlyfacilitates parsing or production of similar or identical syntactic structures(targets)(Pickering&Ferreira,2008).

Recent experimental work within the syntactic priming paradigmsuggests that the effects of priming are not purely syntactic, but are alsoobservable on the level of information structure, and that both, adults andchildren, are susceptible to these effects (Fleischer, Pickering, & McLean,2012; Vasilyeva&Waterfall, 2012). For example, it is claimed that passivesemphasize the patient of the action, and that hearing a passive sentencewouldpromoteproductionofnon-passiveconstructionswhichhavethesame

(π1),causingher fathertorebukeher forspendinghermoneysocarelessly(π2).-Result(π1,π2)

(3) #Mary enjoyed an extremely expensive meal at the Balmoral

(π1), eating a decadent salmon fillet with a lemon and herbsauce (π2) and causingher father to rebukeher for spendinghermoneysocarelessly(π3).- ???Elaboration(π1,π2)&Elaboration (π1,π3)&Result (π2,π3)

(4) Johnbehaved very badly at the birthdayparty, breaking all oftheglassesonthetable.-breaker=John

(5) John behaved very badly at the birthday party, causing hismotheragreatdealofembarrassment.-breaker=John’sbehavingverybadlyatthebirthdayparty

(6) Johnbehaved very badly at the birthdayparty, breaking all oftheglassesonthetableandcausinghismotheragreatdealofembarrassment.-breaker=John&causer=John

15.20-15.45Construction grammar and lexical polysemy: A case study of the verbfeedintwoargumentstructureconstructionsIvanaKráliková•MasarykUniversity

Muchofworkonlexicalpolysemy(Taylor,2012;Cruse,1995;Gibbs&Matlock,1997)hasstressedthattheultimategenerationofmeaninggreatlydepends on features of linguistic context, including collocation and theconstructions words occur in. Exploring the emergence of metaphoricalmeaning1 in different argument structure constructions, Sullivan (2013)argues thatmost verbs arenot fully semantically specified: they relyon thearguments they take for the ultimate generation of meaning. She furtherpoints out that when an argument structure construction is interpreted asmetaphoric, the target domain is evoked by one or more of the verb‘sargumentswhile the verb itself evokes the source domain, thus lacking theabilitytopromptmetaphoricinterpretationonitsown.

FollowingSullivan’s(2013)insights,thispaperexaminestheEnglishverb feed in two different argument structure construction (in sense ofGoldberg, 1995): the transitive construction (e.g. feed the baby) and thetransitiveresultativeconstruction(e.g.feedmilktothebaby),analysingtheirroleintheemergenceofmetaphoricalmeaning.Basedonacorpusanalysisoftheverbinthetwoconstructions,itfocusesoncaseswherethethematicroleof PATIENT is assumed by a word denoting a human entity, such as child,baby,family,peopleorpopulation.ThePATIENTroleisattributeddifferentlywithinthetwoconstructions:itcorrespondstothedirectobjectoffeedwithin

30 19

thetransitiveconstruction(e.g.feedthebaby)andtotheindirectobjectinthetransitiveresultativeconstruction(e.g.feedmilktothebaby).

Preliminary results of the analysis suggest that the differences inthematic role distribution affect the ability of the verb feed to generatemetaphorical meaning. In the transitive construction, the verb feed isinterpreted literally each time it is combined with a human PATIENT. ThesourcedomainofNUTRITIONisevokedbytheverbitselfwhileneitherofitsarguments evokes a different conceptual domain. The THEME is leftunexpressedinthetransitiveconstructionbutisassumedtobefood,possiblyincombinationwithwaterorothersubstancesnecessary forahumanbeingtothrive.

Inthetransitiveresultativeconstruction,however,thedirectobjectposition is occupied by the THEME, which is thus explicitly expressed asopposedtojustbeingassumed.Here,theTHEMEcanbeexpressedbyanounevokingaconceptualdomainotherthanNUTRITION,suchasINFORMATIONTRANSFER in They are feeding lies to the people. In such cases, the entireconstruction, including the verb feed is interpreted metaphorically, asreferringtothetransferofinformationratherthansustenance.Theseresultsseem to indicate that the ability of the verb feed to interact with multipleargument-structureconstructionsystematicallycontributestoitspolysemouspotential by allowing it to specify different thematic roles, some of whichfacilitatemetaphoricalextensionintodifferentconceptualdomains.SESSION8 TUESDAY30THMAY15.55-17.1015.55-16.20AscientificrealiststancetowardssociolinguisticmetatheoryJohannesWoschitz•UniversityofEdinburgh

Allegedly, sociolinguistics as a scientific discipline appearedon theacademicsceneinthe1960s.Focusingontherelationshipbetweenlanguageand society, it sought to leave old linguistic paradigms behindwhile at thesame time establishing its ground alongside the Chomskyan school ofgrammar. To do so, scholars from different backgrounds worked togetherinterdisciplinarily,andoutofthiscollaborationbranchescameintobeingthatnow show signs of incipient separatism, leaving sub-disciplines mainlyconcerned with their own study areas that are still commonly subsumedunderthesametermofsociolinguistics.

Withinthescientificcommunity,thisseparatismhasbeenregardedas given, without any awareness of the possible magnitude and impactongoing philosophical tensions below the surface might have on theirmethodology, modes of data interpretation and linguistic theorising as awhole.Intherecent2016bookSociolinguistics:TheoreticalDebateseditedbyCoupland,however,importantscholarsinthefield(Silverstein,Gal,Bell,etc.)have started toaddress this issuebydebunking trivialities in sociolinguisticfindings, arguing that these can be traced back to a shaky commonphilosophical ground. Turning to the plurality of sociolinguistic disciplines,they weigh their suitability for the volume’s newly postulated goal ofsociolinguistics that should ultimately enable it to progress as a unified

31

Inthisstudy,weadministeredtheAX-Continuous PerformanceTask(AX- CPT) to 201 participants to evaluate the existence of a bilingualadvantage.OriginallydevelopedtomeasurecognitivecontrolprocessesinEFdynamically (Braver et al., 2009), this taskwas recently adopted in a studythatcomparedbilingualandmonolingualperformance(Moralesetal.,2013).Here, we tested Italian-Sardinian bilinguals (46), Italian-English bilinguals(54), Italian-SardinianPassivebilinguals (43)and Italianmonolinguals (58).Whenwecomparedaccuracyasindividuallyaggregatedproportions(i.e.eachparticipanthasasinglescoreforeachtypeoftrial),acrossgroups(controllingfor differences in age and years of education), we replicated the results ofMorales et al. (2013): therewas a significant interaction between conditionand group, with Italian-Sardinian participants performing better than theItalianmonolinguals in theAY condition,whichhasbeen claimed to engageinhibitory control (p = .012). The Behavioural Shift Index, a compoundedscore originally proposed by Braver et al. (2009), also showed betterperformanceinthebilingualparticipantsthaninthemonolinguals(p=.032). However, when we analysed the accuracy data using binomialmixed-model regression, which allows for a more accurate random effectsspecification, we found no evidence for a bilingual advantage: pairwisecomparisonbetweengroupsandconditionsdidnotreachsignificance.Theseresults show how studies investigating the bilingual advantage in EF arelimited in their generalizability. Critically, our study demonstrates theimportanceof factoring in individual variabilitywhenstudyingbilingualism,and it highlights three fundamental problems in current research on therelationbetween languageandattention–namely,a theory-drivenselectionof experimental tasks, the choice of analytical strategies as well as limitedsamplesizes.10.50-11.15Effects of conceptual accessibility and similarity in simple sentenceproductioninMandarinYangziZhou,HollyBranigan,&MartinPickering•UniversityofEdinburgh Mandarinhasthreewaystoproduceasimplesentence:

a)Canonicalactive:(Helen)da-le(Johnny)‘(Helen)hit(Johhny)’b)BA-active:(Helen)BAJohnnyda-le‘(Helen)hitJohnny’c)BEI-passive:(Johhny)BEI(Helen)de-le‘JohnnywashitbyHelen’Bock &Warren (1985) proposed that language production can be

affectedbyentities’conceptualaccessibility,referringtotheeasetoretrieveanentity’s concept from long-term memory. Prat-Sala & Branigan (2000)categorized conceptual accessibility into inherent (intrinsic semanticcharacteristics such as animacy) and derived accessibility (temporaryaccessibility gained by contextualmanipulation such as saliency). Effects ofanimacy received convergent results (e.g. McDonald, Bock & Kelly 1993),however,conflictingevidencewasfoundregardingtheeffectsofsaliency(e.g.Osgood1971;Bock&Irwin1980).Inaddition,conceptualsimilaritybetweentheelementscouldalsoconstrainlanguageproduction(e.g.Gennari,Mirkovic&MacDonald2012).

18

scientificdiscipline:thecontributiontosocialtheory.In this talk, I will engage in the metatheoretical discussion by

rethinkingthiscontributiontosocialtheoryfromanotherperspective.First,Iwill philosophically reframe the three most prominent subdisciplines ofsociolinguistics, namely linguistic variationism, linguistic anthropology andsociologyoflanguage.Iwillfocusontheformertwoastheseareconsideredto be antagonists in contemporary metatheoretical debate. Subsequently, Iwillshowhowsociolinguisticinterestshavechangedrecentlybypresentingashort content analysis of contemporary sociolinguistic scholarship, andwillargue that the above-mentioned distinction of sociolinguistic subdisciplinesno longerdoes justice to the complexityof sociolinguistic explananda. Iwillbriefly summarise how the contributors to the Coupland (2016) volumereflectonthischange,andwillgoontoarguethattheirratherprogrammaticstance can be underpinned by drawing from the scientific realism/meta-inductive pessimism debate within the philosophy of science (cf. Ladyman2016).Iproposethatfromtheviewpointofascientificrealist,thetransitionawayfromclassicalvariationismtowardsanall-encompassingsocio-semiotictheory of language is a sign of a maturing science. This maturation can beexplained by the grade of self-awareness of value-laden presumptions andexplanatory value, which both depend on the degree of ontological self-justification.16.20-16.45TheUnitedStatesof‘Merica:IndexicalityandTwitterHashtagUseRachelMoyer•UniversityofEdinburgh InaheatedtimeintheUnitedStates,socialmediaandmicrobloggingplatforms, likeTwitter,havebeenusedasforumstoexpresspoliticalbeliefs,social critiques and frustrations deriving from the cultural constructs inAmericansociety.Exploringthemediaplatformswherepeopleexpresstheirexasperation for thecurrent social climate, cangive insight intohowpeopleindex the society inwhich they live. Recent sociolinguistic research (Shapp2014; Page 2012; Papacharissi 2012; Chang 2010) has discussed how thesocial has manifested in Twitter hashtag use, finding that hashtags havedeveloped past their original organizational function and now contributeadditional meaning to the message. Hashtags can be used in many socialprocesses like self-evaluation, image-building and self-presentation. Newstudieslinkingresearchbetweenthesociolinguisticsofmicrobloggingsitestoindexicaltheoryarestillcalledfor.

Indexical research has discussed the importance of context inattributing meaning to a specific linguistic norm (Jaffe 2016; Eckert 1989,2008;Gal1978,Silverstein2003).Byintroducingtheideaofindexicalorders(Silverstein 2003), macrosocial theories meet microsocial practice.Indexicality emphasizes that the relationship among indexes is a cyclicalprocess of reinterpretation (Eckert 2008, 463). Explicating the network ofindexesinanindexicalfieldallowsindexestobeinterpretedinrelationtooneanother. Using indexes to create a context for hashtags, indexicality meetssocialmediainthispaper.

The#MericatrendonTwitterismostexemplaryofthepoliticalandsocialschismsintheUS.Thispaperanalyzestheindexicalfieldof‘Merica’in

older forms for third-person singular possessive (particularly thereof). Thelinguistic evidence may illustrate the strong conformity of Puritans to theWordsofGod,thetraditionofEnglishvernacularBibleswhichretainarchaicformsoflanguage.Thesamecorrelationhasbeenobservedintheuseofthird-person singular inflection (-th vs. -s)which have been reported in previousconference presentations; thus, I will attempt to generalize the overallrelationshipbetweenreligiousidentityandlanguagepractice.9.50-10.15DiscourseandtopicalizationinoldEnglishsubordinationSergioLópezMartínez•UniversityofOviedo Thecloserelationbetweendiscourseandsyntax inOldEnglishhasbeen a favourite topic for research over the last few years, as seen, forexample, in Kemenade and Los (2006), Kemenade (2009), or Kemenade &Milicev(2011).However,mostoftheexistentworkontheinterplaybetweeninformation structure and syntax inOld English focuses onmain sentences,andtheinstancesinwhichsubordinatesentencesarestudied,itisinrelationtothediscourse-relatedparticlesþaandþonne. Inpreviouswork(LopezMartınez2016),IprovideddatasupportingtheclaimthatembeddedtopicalizationispossibleinOldEnglishsubordinatesentences, and it was suggested that this phenomenon could be related todiscourse factors, in connection to van Bergen (2015). Thus, the aim of thepresentpaper is tostudyhowdiscourse factorsmay influencetopicalizationinsubordinateclauses.Inordertodoso,alargecorpusofprosetextsfromtheOE period will be analysed, examining not only the instances of embeddedtopicalization insubordinatesentences,butalsohowdiscourseaffects thosetopicalizedelements.Thisstudywillpayattentiontowhetherthoseelementsarethematic(i.e.giveninformation)orrhematic(newinformation),analysingthe discourse preceding the relevant subordinate clauses in relation to thecurrenttheoriesonInformationStructure.SESSION5 TUESDAY30THMAY,10.25-12.0510.25-10.50Mechanisms of cognitive control in bilinguals: The empiricalgeneralizabilityofthe‘bilingualadvantage’MichelaBonfieni•UniversityofEdinburgh The relation between the bilingual linguistic experience andcognitive control is hotly contested. Several studieshave examinedwhetherthe bilingual linguistic experience is beneficial to Executive Functions (EF),withdifferentexperimentalparadigmsandinconclusiveresults.Themajorityof those studies tended to focuson cognitive control as a singlemechanismrather than as a multitude of processes dynamically related to each other(Braver 2012; Green and Abutalebi, 2013), and used tasks that yieldeddivergent measures of EF (Paap and Sawi, 2014). Moreover, small samplesizes and non-conservative analytical methods put into question thegeneralizabilityofsuchresults.

32 17

some basic descriptive statistics. Secondly, I present the results of ageneralised linear model (GLM) fit to the data, showing how this canprovidefurtherinsightintotrendscomparedwithbasicstatisticalmethods.Lastly,Ipresentageneralisedadditivemodel(GAM)fittothesamedataset,againshowinghowthismethodimprovesinsightintothedata.

9.25-9.50Language, identity and community: The use of third-person neuterpossessivesofearlymodernpreachersHiroshiYadomi•UniversityofGlasgow This paper examines the use of third-person neuter singularpossessive pronouns (its, his, of it, and thereof) of 20 Early Modernpreachers. In theperiodexplored, thenewformitswasnewly introducedand other variants were also available to express the same meaning(Nevalainen & Raumolin-Brunberg 1994; Barber 1997; and Lass 1999). Iwillshowthatthereisahugevariationinthechoiceofthird-personneuterpossessives in preachers’ language and discuss possible factorsconditioningthevariationbetweenindividuals.Inadditiontolinguisticandmicrosociolinguistic factors (sex,ageandclass), Iwill shed lightonmoremicro sociolinguistic parameter of sermonwriters, namely their religiousidentity.ThisisapartofmyPhDprojectanalysingthecorrelationbetweenEarlyModernpreachers’religiousidentityandtheirlanguagepractices.Theproject aims to describe how individuals and communities interact withongoinglanguagechanges. The data used in this study is retrieved from the Corpus ofSermons in Early Modern England, which has been compiled to analyselinguistic variation between individual speakers in the sole genre ofsermons (Yadomi 2016). The corpus contains one millions words, with50,000 words from sermons ascribed to each of the 20 preachers. Thesermons included in the corpuswere published between 1588 and 1660,duringwhich the third-person neuter possessive pronouns experienced ahuge,rapidshift. The study will focus on the relationship between preachers’languageuse and their religious identity, towhicha community isdeeplyrelated.EarlyModernpreachersformadiscoursecommunitywhichshares“common goals” and “linguistic practices” (cf. Swales 1990) as well asprofessionalidentity.However,thecommunityisbynomeanshomogenousin that therewas a rather clear divide amongmembers in terms of their religious identity, which lies subordinate to the professional identity(Kiesling 2013). In any sort of community, there arises voluntary orcompulsory social force which accommodates its members to thecommunitynorm.TheconfessionalstatusofEarlyModernEnglandcanbeclassified to Anglican conformist, moderate Puritanism and radicalPuritanism(Lake1988:7). In the presentation, I will discuss how ideologically conflictingparties within the discourse community had different preference inlanguage choice. Pilot studies show that Puritan preachers tend to retain

28

33

Twitter hashtag use and the rhematization process it undergoes. UsingTwitter extracts as the primary data, this paper first looks at how hyper-American concepts index ‘Merica,’ then goes on to assert that ‘Merica’ hasdeveloped a natural relationship as an icon for the heightened, polarizingcliches of American society. With reflexivity being at the heart of indexicalinterpretation,theimportanceofcontextisreinforcedinthispaper.Lifestyleindexes provide the context needed for other ‘Merica’ indexes to take on areflexiveform.Withsuchacomplex,richfeature,thisanalysisemphasizesthescale of reflexivity seen among indexes. This paper suggests that speakersdrawon‘Merica’whenitsupportstheirpoliticalorsocialcritiqueinTwitterpublications. With its emphasis on indexicality, this paper calls for furtherexplorationintohowothersocialmeaningislinguisticallygivento‘Merica.’ 16.45-17.10ThechangingSwedish/iː/vowel:Evidence fromthreeCentralSwedishcitiesFabienneWesterberg•UniversityofGlasgow

TheCentralSwedish/iː/vowelistraditionallyrealised[iː],buttoan

increasing extent, speakers are beginning to use another variant known asViby-i. This is a vowel with an unusual “thick”, “dark”, “buzzing” quality(Borgström,1913)foundin“severalscattereddialects[acrossSweden],bothin rural areas and in the city dialects of Stockholm and … Gothenburg”(Björsten & Engstrand, 1999). Viby-i has not been extensively studied, andthere aremany questions surrounding e.g. how its unusual vowel quality iscreated,whetheritisproduceddifferentlyindifferentlocations,orevenhowfaracrossSwedenithasspread.

This paper presents an acoustic apparent-time analysis of a smallsample of speakers from Stockholm, Gothenburg and Uppsala in CentralSweden (Fig 1). Only the first two cities are known to have Viby-i. InStockholmitseemstobearelativelynewfeature,whichwasfirstdocumentedinthe1990s(Kotsinas,1994),andthenprimarilyusedbyyoungupper-classfemales.Sincethisgroupisoftenattheforefrontoflanguagechange(Trudgill,1974), it is likelythatthefeaturehascontinuedtospreadwithinStockholm,but that it is most common in young and middle-aged speakers. InGothenburg,Viby-i isconsiderablyolder,withearlyrecordsdatingitbackto

16

34 15

learner’s proficiency and but also reading proficiency. Findings first indicatedthat participants with lower reading fluency improved their readingcomprehensionthroughthisexperiment.Second,it showedthatoralreadingiseffectiveinpromotingautomaticphonologicalcodingandacceleratingthespeedofvocalspeech.Theresultssupportapedagogicalimplicationthatoralreadinghas an instructional value to improve learners’ speaking and pronunciationalfluencyaswellasreadingfluency.SESSION4 TUESDAY30THMAY,09.00-10.15 9.00-9.25ScarystatisticsandhistoricallinguisticsDaisySmith•UniversityofEdinburgh

Theavailabilityof large-scalediachroniccorporahasmadeitpossibleto investigate historical linguistic phenomena using sophisticated statisticalmethodology. Gries (2015, 97) describes corpus data in general as“observational and, thus, usually unbalanced andmessy/noisy”. He points outthat techniques frequently employed in other areas of linguistics, particularlythose which use experimental methods, can be usefully applied to corpuslinguistics. More than that, he suggests that these methods are activelynecessitatedbytheverynatureofcorpusdata.

Historicalcorpusdata isnoexceptiontothisgeneralisation,butHistoricalLinguistics is a field inwhich the scope for ‘bigdata’ analysishasbeenunder-standablysmallerthaninothers.Compare,forexample,thefollowingcorpora:

1. TheEdinburghTwitterCorpus(Petrovic,Osborne,&Lavrenko,2010):acorpusofsocialmediainteractionscontaining97milliontweetswithacom-binedtotalofover2billionwords.

2. ALinguisticAtlasofEarlyMiddleEnglish(LAEME)(Laing,2013):acor-pus of Middle English (ME) texts containing approximately 300transcribed Medieval manuscripts with a combined total ofapproximately650,000words.

Petrovic et al.’s corpuswas extracted automatically over a period of

twomonths using Twitter’s streaming API. By contrast, LAEMEwas compiledmanually over the course of 20 years by the manual sourcing, reading,decipheringandtranscribingofmanuscripts(Laing,2013).Evenconsideringtherelative length of the LAEME texts compared with 140-character-maximumtweets, these two corpora are clearly on vastly different scales. Where theelectronic data from Twitter can be extracted automatically, the LAEME data,and all similar historical corpus data, must be painstakingly transcribed anddigitised.

How,then,inafieldwherehand-drawnisoglossesareformanyanot-so- distant memory, should a researcher embarking on a complex statisticalanalysispresenttheirresultstothehistoricallinguisticcommunity?

In this paper, I present an analysis of data fromA Linguistic Atlas ofOlder Scots (LAOS), firstly considering what information can be gleaned from

at least the 1950s (Björseth, 1958). It might therefore be expected to bepresent inolderGothenburg speechaswell. InUppsala,Viby-ihasnotbeenrecorded,butitisstillpossiblethatithasfounditswaytherefromStockholm,whichisonlyanhour'sdriveaway.Uppsala ishoweverregardedasclosetothenationalstandarddialect(Källskogetal,1993),whichmeansthatitmightalsoprovideagoodreferencepointforstandard[iː].

The aim of the paper is thus to investigate how far along the shifttowardsViby-ihascome,primarilybylookingatageandlocation.However,IamalsointerestedinthedistributionofViby-iacrossothersocialfactors,suchasgenderandsocialclass,sincethismighttellusbothaboutitsspreadanditssocial value. In urban areas in particular, Viby-i is associated with highprestige(Bruce,2010),whichmaymakeitmorecommonincertaingroups.Itisalsopossiblethatspeakersproducethevoweldifferentlybasedonsocialorgeographicalfactors.

Theresearchispartofalargerprojectinvestigatingtherelationshipbetweenarticulation,acousticsandsociophoneticsintheproductionofViby-i.Thesubsamplepresentedhereconsistsof18femalespeakersaged20-25and50-80producingphrase-listutterances,withapproximately50/iː/tokensperperson. The remaining long Swedish vowels /yː, ʉː, eː, øː, ɛː, ɑː, ɔː, uː/ areincluded for context. The acoustic analysis is primarily based on formantvaluesforF1-F3.

SESSION9 WEDNESDAY31STMAY,9.00-10.1509.00-09.25Linguisticnicheoracademicniche?Theroleoflanguage-externalbiasesinlanguageevolutionJonasNölle•UniversityofEdinburgh Over the past decade, a number of major claims have been madeabout the fundamental nature of human language(s). For instance, from anevolutionaryperspective,ithasbeenarguedthatthestructureoflanguagesisnot theoutcomeofasinglebiologicaladaptation,assomewouldargue(e.g.,Berwick & Chomsky, 2017), but rather the result of biological-culturalinteraction (e.g. Christiansen & Chater, 2008; Kirby 2016), which – overhistoricaltimescales–gaverisetostructureanddiversity(Evans&Levinson,2009). On such a view, language is a complex adaptive system (CAS, seeBeckneretal.,2009)thatconstantlyevolvesasitislearnedandusedwithinapopulationofagents.Morerecently,thisviewhasbeencomplementedbytheidea of linguistic adaptation (herafter LA, see Lupyan & Dale, 2016 for areview). The hypothesis is that languages as CAS tend to adapt to theirphysical,socialortechnologicalenvironment,similarlytoorganismsadaptingto their local ecological niche. This has been supported by a number ofcorrelations that have been found in recent years, e.g., the relationshipsbetween morphological complexity and population size (Lupyan & Dale,2010),linguistictoneandhumidity(Everettetal.,2015),lexiconandclimate(Brown & Lindsey, 2004; Regier, Carstensen & Kemp, 2016) as well asanatomicalandgeneticbiases(Dediu&Moisik,2016;Dediu&Ladd,2007).

Inspired by Gupta and colleagues’ (2017) critique of niche

35

construction in evolutionary theory, this paper tries to critically assess towhat extent such correlational claims can be meaningfully integrated incurrent accounts of language evolution andwhether LA should be added asone of themajor driving forces underlying linguistic structure. In line withRoberts & Winters (2013), I argue that correlational studies like the onesabovecangeneratehypotheses,butshouldbeviewedwithcaution,sincetheycannotprovidecausalexplanations.Inordertoaddresstheroleoflanguage-external pressures, I suggest to a) utilize such large-scalestatistical/correlational studies to formulate questions about possibleexplanationsfor linguisticdiversityandb)test thesehypotheses incarefullycontrolled laboratory experiments. These allow isolating mechanismsresulting in structural patterns that resemble distributions we can observeamongthelanguagesoftheworld.AsanexampleforhowtheorieslikeLAcanbenefit from the integration of complementary empirical approaches, I willpresent data from two experiments. The first one had dyads solve acoordinativemazegame,wheremanipulationoftheshared-taskenvironmentcould motivate different linguistic strategies based on differentconceptualisations of the otherwise identical task. The second experiment,relying on the iterated learning paradigm (Kirby, Griffiths & Smith, 2014),investigatedhowusagecontextcaninfluencetheemergenceandevolutionofoverspecificationofasemanticmarkerinanartificiallanguage(Tinits,Nolle&Hartmann, 2017, in press). In sum, these studies show how hypothesesconcerningLAcanbeformulatedandtestedbyintegratingcorrelationalandexperimentalwork.

09.25-09.50Theculturalco-evolutionoflanguageandmindreadingMarieke Woensdregt, Simon Kirby, Chris Cummins, & Kenny Smith •UniversityofEdinburgh Language use requires an understanding of communicativeintentions,whichinturnreliesontheabilitytorepresentother’sminds(e.g.Moore,2016).Thedevelopmentof suchmindreading skills formeda crucialstep in hominin evolution, because it allowed for the expression andrecognition of communicative intentions, thereby paving the way for thecooperative information sharing we find in humans today. The ability torecogniseandinfercommunicativeintentionsalsoplaysanimportantroleinlanguagedevelopment,asevidencedbystudiescorrelatingmindreadingskillsand word learning (e.g. Parish-Morris, Hennon, Hirsh-Pasek, Golinkoff, &Tager-Flusberg,2007).Thisrelationshipbetweenlan-guageandmindreadingmay be reciprocal; the acquisition of language has been shown to unlockfurtherlevelsofmindreadingdevelopmentintheindividual(e.g.Lohmann&Tomasello, 2003; Pyers & Senghas, 2009). Furthermore, Heyes and Frith(2014)arguethatthesophisticatedmindreadingskillsofmodern-dayhumansareinpartaresultofcumulativeculturalevolutionwhichwaslikelyaidedbylanguage.

Inthispaperwepresentanagent-basedmodelthatinvestigatestheimplica- tions of such a bidirectional interaction between language andmindreading.Inthismodel,mentalstatesareimplementedastheproductofaninteractionbetweenanagent’s‘perspective’(theirviewontheworld)and

14

(WCF) facilitates learners’ development in a second language (L2), and, if so,whethermoreorlessexplicittypesofWCFaremoreeffective.Researchersarekeen todiscoverhow learnersprocessWCFand tounderstandhow,whyandwhendifferentstrategiesofWCFfacilitateL2development. The research project described in this talk investigates theeffectiveness of two types of WCF (direct and indirect) in improvinggrammatical accuracy of 77 Omani EFL students on two linguistic structures(comparativesandprepositionsofspace)bymeansofpre-test,immediateposttestanddelayedposttest. Thinkaloudprotocolswerealsousedwith6studentsfromeachgroupto investigate how students process the feedback given (direct vs. indirect)whiledoing revisionsandnew tasks.Thispresentationwill focuson the thinkaloudprotocols.Drawingondatafromthestudy,Iwillshowthatbothcognitivefactors and my own scaffolding approach influenced the way that learnersprocessedtheWCF.Furthermore,students'reportsrevealedinterestingfindingsrelatingtohowdifferentindividualsprocessedtheWCF. Findingsshowthatgreateraccuracyonpost-testsresultedwhenWCFwasnegotiated.ThisresultchimeswithNassaji'sstudy(2012)whichfoundthatnegotiated feedback ledtogreateraccuracyonpost tests thannon-negotiatedandminimallynegotiatedfeedbackfortheuseofarticles.18.05-18.30TheinfluenceoforalreadingonKoreanlanguageproficiency:AcasestudyJihyeEo•YonseiUniversity Thisstudywasdesignedtoinvestigatetheeffectoforalreadingon26collegestudentsofKoreanasaforeignlanguage.Traditionally,oralreadinghasbeen known teaching methods to teach a foreign language, especially fordevelopingreadingfluency.Thesedays,oralreadingthatwentunnoticedintheprevious researches is now being revalued as a significant skill to improveforeign language learners’ communicative competence.Alongwith this changeto new awareness and evaluation of oral reading, there have been severalresearchesthatstudytherelationshipbetweenoralreadingandotherlanguageskills. ThestudywasconductedfromtwodifferentKoreanlanguagecoursesataprivateuniversityintheUnitedStatesin2014and2015.Intheselanguagecourses,whereheritageandnon-heritagelearnersofKoreanlearnedthetargetlanguage together, total14 students enrolled in “IntermediateKorean” and12students enrolled in another “IntermediateKorean”. The researcher itselfwastheinstructorofthesetwoclasses,whichconsistedoffour50-minutemeetingsforthedurationofa15-weekssemester.Thestudentswereaskedtoreadaloud26passagesandrecord themselvesovera13weekperiod.Eachrecodingwasreviewedandcommentedbytheresearcherwhothenratedeachrecordingonanumerical scale. Five perspectiveswere used to reviewed and commented forthe participants; (1) accuracy of the word pronunciation, (2) expression andvolume, (3) phrasing, (4) pace and (5) prosody, and all performance weregradedona1 to4 scalewith4being thehighest.And the testof readingandspeaking testwas carried out oneweek after the performance of oral readingwasconducted.

Theresults in thisstudyshowedthatoralreadingeffectsonnotonly

35

thecontext(thestateoftheworlditself).Perspectiveandcontextinteracttoproduce the agent’s mental state, which influences how likely they are tochoose the different potential referents as their communicative intention.Whiletheagent’sperspective is fixed, thecontext ischangeable;renderingadifferentmentalstate foreach interaction. Inadditiontoaperspective,eachagenthasalexicon(mappingofsignalstoreferents),whichtogetherwiththecommunicativeintentionproducesutterances.

Duringdevelopment,eachnewagentobservesasetofcontextsandcorresponding utterances from their cultural parent. From these twoobservable variables the learner has to infer two unobservable variablessimultaneously: the parent’s lexicon and perspective, using Bayesianinference.Thedevelopmentalresultsshowthatthisjointinferencetaskleadsto co-development. A parent’s lexicon can only be inferred correctly if thelearner can represent the possibility of the parent having a differentperspectivefromtheirown,andtheperspectivecanonlybeinferredcorrectlyif the parent uses an informative language. Thus, learning of lexicon andperspectiveareinterdependentinthismodel.

This leads to the question of under which circumstances apopulation of such agents is able to establish a conventional language fromscratch,i.e.ifallagentsinthefirstgenerationuseacompletelyuninformativelexicon. The evolutionary results show thatwithout any additional selectivepressure, an informative language does not evolve. Selection for successfulcommunication can provide the necessary pressure to lead populationstowards convergence on an optimal language. Interestingly however,selectiononperspective-takingcan leadpopulations to thebeginningsofaninformative system as well; stopping at whichever point the language isinformative enough to learn perspectives correctly given the number ofobservationsthatthelearnersreceivefromtheirculturalparent.Thismodelthusgivesinsightintotheeffectsofanindividual-levelinteractionofcognitivecapacitiesonpopulation-widedynamicssuchasestablishingandmaintainingastablesignallingsystem;therebyconnectingproximateandultimatecausesoflanguageevolution.

09.50-10.15Investigating effects of morphological types and their word formationpatternsinSLASvenjaWagner•UniversityofEdinburgh Previous research in SLA has shown that knowledge of structuresthroughthemothertonguecansupporttheacquisitionofsimilarstructuresinan L2 (Gass & Selinker, 2008). At the same time, factors such as semanticcomplexity and morphophonological regularity have been shown to impactacquisition ease also independent from the learner’s mother tongue(Goldschneider & DeKeyser, 2001). My research combines both theoreticalapproaches and investigates to which extent morphological language typescan predict learners’ success, both in relation to and independent from thetypeofthelearner’smothertongue.Intwoartificiallanguagelearningstudies,I have modeled three morphological types, isolating, inflectional, andagglutinating, based on their mapping of grammatical meaning ontomorphemes and the combination of morphemes to form complex words.

36 13

ofmetadiscourseinEGAPclassesbyEFLteachersacrossChinaandUK.Inlightof Hyland’s (2005) interpersonal model and Adel's (2010) reflexivemodel ofmetadiscourse, a revised model is to be proposed, and the two corpora arecompared to analyse the differences between their metadiscourse uses ingeneral,andindifferentmodesofclassroomteachingactivitiesbasedonWalsh(2006, 2011, 2013) in particular, and to explore the influence of pedagogicalcontent and context on teachers’ choice of metadiscourse in classrooms. Thisstudy also has further implication on EFL teachers’ training and developmentprogram. SESSION3 MONDAY29THMAY,17.15-18.3017.15-17.40Across-linguisticapproachtoadpositionsinthirdlanguageacquisitionSakineÇabuk•MiddleEastTechnicalUniversity This study explores the role of cross-linguistic influence in thirdlanguageacquisitionprocessbyexaminingEnglishadpositions.Comprehension,processing and production of English prepositions (in, on, at, behind, over, to)were examined through off-line and on-line data collection tasks to find outwhich of the two known languages (L1 or L2) is the major source of cross-linguisticinfluenceontheacquisitionofEnglish(L3)adpositionsgiventhefactthat adpositions are morphologically and syntactically different in Turkish,Kurdish and English languages. The main reason behind the choice of theseparticular prepositions lies in their morpho-syntactic properties in Kurdish,Turkish and English. While some adpositions have similar representations inthese languages (e.g., behind, overappear as prepositions in bothKurdish andEnglish), some others have different representations (e.g., in, on, at appear asprepositions in English, case markers and/or postpositions in Turkish andprepositionand/orcircumpositionsinKurdish).ParticipantsofthestudywereTurkish-Kurdish bilinguals who formed experimental group and L1-Turkishmonolinguals who served as control group. Two off-line picture descriptiontasks (picture description task with multiple choices and teddy bear picturedescription task) and an on-line self-paced reading task were employed tocollect data. The finding of the study revealed that Turkish-Kurdish bilingualswere better in comprehending, producing and processing target prepositionsthanL1-Turkishcontrolgroup,particularlywhentheyhavestructuraloverlapsbetween the adpositional systems of L1-Kurdish and L3-English. The findingsaresuggestiveoftypologyasanoverridingfactorincross-linguisticinfluenceintheacquisitionofL3English.StructuraloverlapsbetweenKurdishandEnglishfacilitatedtheacquisitionofEnglishprepositionforTurkish-Kurdishbilinguals17.40-18.05Think aloud protocols: A window into the minds of how English as aforeignlanguagestudentsprocessfeedbackSharifaAlHarrasi•UniversityofStirling Therehaslongbeenadebateonwhetherwrittencorrectivefeedback

Dependentontype, languagescouldallowa fusionofgrammaticalmeaningsin single morphemes (semantic complexity), draw on free and boundmorphemes (syntactic category) to various extents, and exhibit variants ofmorphemes based on their phonological environment (morphophonologicalregularity). Differing levels of these predictors in morphemes have had astatisticallysignificanteffectonthesuccessoflearningthemorphemesinthestudies.

Inthefirststudy,conductedwithnativespeakersofthreedifferentlanguages,themothertongueofthelearnershowedaneffect,butislikelytobe correlatedwith cultural factors.The language typeof theL2, in contrast,proved to be a reliable predictor of learning difficulty, even independent ofthe learner’s mother tongue. Further, the isolating artificial language,exhibitinga largeamountof freemorphemes,whichwerealsonon-complexintheirstructure,waslearnedsignificantlybetterthantheothertwotypes.Atanearlystageoflearning,semanticcomplexityappearedtoinhibitsuccessofacquisitionlessthanmultimorphemicwordswithanunderlyinglesscomplexstructure. These findings emphasize the role of visually salient freemorphemes and they allow for the careful assumption that facilitatedwordsegmentation due to a low level of semantic complexity did not contributemuchtoeaseoflearning.

In the second study, conducted with English native speakers,however, low semantic complexity did facilitate ease of learning incomparison to a conditionwith semanticallymore complexwords.Due to adifference in word length between both studies, the findings suggest theexistence of a threshold at which clarity of word structure does no longercounterbalancethedifficultyresultingfromwordlength.Overall,thefindingsfrom both studies show that certain word structures are easier to acquirethan others, independent of the learner’s mother tongue, and consequentlysupporttypologicallybasedaccountsoflearningdifficulty.

SESSION10 WEDNESDAY31STMAY,10.25-11.1510.25-10.50The development of tense morphology by Welsh-English bilingualchildrenwithandwithoutSpecificLanguageImpairment(SLI)HyowonKwon&VickyChondrogianni•UniversityofEdinburgh Todate,muchresearchconcerningtheacquisitionofEnglishtense-marking morphology as a clinical marker has focused on comparingmonolingual childrenwith SLI to theirTDpeers (Leonard, 1998;Rice et al.,2000;Rice&Wexler,2001;Bishopetal.,2006).StudiescontrastingchildrenwithandwithoutSLIacquiringEnglishasasecondlanguage(L2)haveshownthatL2childrenhavemoredifficultieswith tensemorphology inproductionthanmonolinguals,andEnglishL2childrenwithSLIshowexceptionaldeficitsin tense morphology compared to their L2-TD peers (Chondrogianni &Marinis, 2011; Paradis, 2005; Paradis et al., 2008). Therefore, looking ataccuracy and error types of tense morphology by L2 children with andwithoutSLIcouldfunctionasthepotentialclinicalmarkerthatdistinguishesTDfromSLIinEnglishL2children(Blom&Paradis,2013;Paradis,2008).Toourknowledge,researchtappingintobilingualchildrenwithandwithoutSLI

12 37

16.15-16.40Emerging self-identities and emotions: A qualitative case study of tenSaudistudentsintheUKOunAlmesaar•UniversityofEssex

ThistalkaimstouncovertheprocessesofL2identityconstructionanddevelopmentamongSaudiArabianstudents in theUnitedKingdom.Guidedbythetheoryofpossibleselves (Markus&Nurius,1986), thetheoriesofsituatedlearning(Lave&Wenger,1991)andimaginedcommunities(Norton,2000),thispresentation will examine the participants’ past and present learningexperiencesandemotions,andtheireffectsonparticipants’orientationandself-perception to English writing. Dörnyei’s (2005) notion of the Ideal L2 self isutilised toexplore identity andSaudi students’ viewsof themselvesaswritersovertimeandinvariouscontexts.Thisexploratorycasestudyemployedvariousinstrumentstocollectdata includingasurvey, twosemi-structured interviews,journals, think aloud protocols, and writing samples. Thematic and narrativeanalyseswereutilisedtoanalysethedata. The preliminary findings show that all participants had a negativelearningexperienceinSaudiArabia;theywereunmotivatedandlackedchancestopracticeEnglishparticularly inwritten form.The learningexperiencehadalittleeffectonsomeparticipantsandnoeffectonothers in theconstructionofpossibleselves.IntheUK,allparticipantshadapositivelearningexperienceandtheirlanguageskillsimprovedsignificantly.TheirL2learningexperienceplayedahugerole intheconstructionofpossibleselvesaswellas inreactivatingandcreating desires in becoming better and more successful writers for mostparticipants. It also helped in sustaining and reinforcing participants’ L2possible selves through creating chances for the participants to be active andautonomous learners. Furthermore, participants’ future possible selves wereperpetuated via joining variousEnglish language courses,master courses, andspendingalongtimeintheUK.TheirviewsofthemselveskeptonshiftingandchangingduetotheiracademiccircumstancesandemotionswhichaffectedthewayparticipantsviewedthemselvesaslanguageusersandEnglishwriters.16.40-17.05Metadiscourseintheclassroom:Acomparativeanalysisofnativeandnon-nativeEFLteachersXinxinWu•UniversityofAberdeen Metadiscourse, as a rhetorical and pragmatic strategy used toconstruct discourse and express the speaker’s or writer’s attitude, has greatpedagogical implications for the classroom teaching of English as a foreignlanguage (EFL) teachers, especially for the non-native EFL teachers whoseclassroomdiscourseformsthemediumaswellastheinputfortheEFLstudents.Researchintometadiscourse,however,hasbeenprimarily focusedonthefieldofwrittendiscoursesuchasdiscourseanalysis,pragmaticstudiesandlanguageeducation,whilethespokenacademicEnglish,especiallycomparativestudiesinteachers’ classroomdiscourse has remained largely unexplored. Based on twocorpora of EFL teachers’ English for general academic purposes (EGAP)classroomdiscourses:5EGAPclassesfromUKuniversitiesand5EGAPclassesfromChineseuniversities,thisexploratorystudyattemptstoinvestigatetheuse

11

SESSION2 MONDAY29THMAY,15.50-17.0515.50-16.15Practicedlanguageattitudes:AcaseofmothertongueeducationinTaiwanChia-YingYang•UniversityofEdinburgh

Taiwan is a multilingual country where Taiwanese Mandarin is thenational language, with Taiwanese (Daighi), Hakka, and 10 Austronesianlanguages as generational or regional languages throughout the country.Moreover, the global Lingua Franca: English (see Seidlhofer, 2005; Graddol,2007; Abdullah and Chaudhary, 2012; Jenkins & Leung, 2013), also plays animportant role in the Taiwanese linguistic repertoire, as it is one of themainstream language subjects in the National Curriculum. Attitudes towardthese languagesconstituteawelldefined linguistichierarchy:Englishplacedatthetop,followedbyTaiwaneseMandarin,thenDaighi,HakkaandAustronesianlanguages(Hong,2002).Suchlinguistichierarchyhastriggeredalanguageshift:theyoungergenerations(theunder30s)ofwhicheverethnicgroupstheybelongto, have broadly switched to become monolingual in Taiwanese Mandarin(Census,2010).

ThisstudyfocusesonDaighiteachers,whoaremultilingualspeakersofat least three languages: Daighi, Taiwanese Mandarin and English, andinvestigatesthenotionofDaighiteachers’useof languageandtranslanguagingduring the interview and in the classrooms. As interviews are my main datacollectionmethod,Iinterviewed20teachers:10teachersteachinginthecapitalcity of Taiwan, Taipei, where based on the census 2010, intergenerationallanguageshiftfromDaighitoTaiwaneseMandarinisshowntobeatthefastestrate compared to the rest of the cities, as Taiwanese Mandarin is thepredominant language (see Yeh, Chan & Chen); and another 10 teachersteaching in Changhua,where the 2010 census indicates over 96% of residentreported using Daighi at home. I also observed two classes per teacher as asupportingdatacollectionmeans.

Arichsetoftranslanguagingdatawasfound.Inthispaper, I focusondiscussingwhatmaypotentially contribute to theoutcomeofonly4outof10DaighiteachersusingDaighiasthecodeofinterview.Ialsoexaminethefunctionofdifferentlanguageswhenusedininterviewsandclassroominstructions.Theresults show that language choices aredependent on the interlocutor, such asteachers’perceptionoftheirinterlocutor’sabilityinparticularlanguages,aswellas a form for the teachers to show their identity and professionalism.Translanguaging,ontheotherhand,isobservedtobeaformofcommunication,aspedagogy,asanactiontopromotelanguageequality,andaslanguageabilitythat teachers expect students to have to prepare them for the multilingualcommunity.Thispaperwilldrawonthedifferenttranslanguagingaspectsintheform of conversation analysis to show detailed language exchange, it alsodiscusseshowsuchtranslanguagingpracticeachieves thepurposesmentionedabove.

has mainly been limited to investigating groups of either younger or olderchildren, or specifically their early and late years of L2 exposure. However,comparingaccuracyanderrortypes intense-markingmorphologyproducedbybothyoungerandoldergroupsofL2-TDchildrenisessentialforrevealingageeffects in theirdevelopmentalpatternsof tensemorphology.Hence,ouraim is to provide a cross-sectional study, which investigates how tensemorphology develops among three different profiles of bilingual childrenacquiring English as L2 across a wide range of ages, and whether thedevelopmentaltrajectoriesoftheir3SG–sandpasttenseacquisitionprofilesareinlinewithpreviousstudiesthatsupportthepredictionofBybee’s(1995,2001) usage-based network model (Blom & Paradis, 2013; Paradis et al,2012).AgroupofWelsh-EnglishTDbilingualchildrenfrom7-9-yearsofage,ayoungergroupofL2-TDchildrenfrom4-6-yearsandanage-matchedgroupofL2-SLI peers were administered the tense probe from the Test of EarlyGrammatical Impairment (Rice & Wexler, 2001). Responses that had beentranscribed and scored on the TEGI were selected and coded in order toanalyzeindividualtensemorphologyelicitationprobes.Ouranalysisincludes1)ageeffectsofL2-TDchildrenontheiraccuracyanderrortypechangesovertime and 2) mixed linear regression to assess whether phonologicalsensitivity,wordfrequency,andL1transferhaveaninfluenceonperformanceof L2-TD children in the early or later stages of development, or acrossdevelopmental stages. Based on findings of howdevelopment patterns varybetweenyoungerandolderL2-TDchildren,asubstudywillcomparetheL2children with SLI to see how they develop tense morphology compared totheir age-matched TD controls. Results predict that tense morphology usecouldpotentiallyrevealdevelopmentalstagesasakey tounderstandingageeffects and profile differences across groups of bilingual children with andwithoutSLI.10.50-11.15Instructedvs.uninstructedbilinguals:TheroleplayedbymetalinguisticawarenessinthirdlanguageacquisitionFrancescaD’Angelo•UniversityofEdinburgh

The present work explores the relatively new field of research

investigatingThirdLanguageAcquisition(TLA),whichconnectstwodomainsthathave traditionallyoverlookedone another: second language acquisitionandbilingualism.

The expansion of research intomultilingualism in the past decadehas been the result of a growing awareness that acquiring a third language(L3) isadifferent,morecomplex,andmultifacetedprocesscomparedto theacquisition of a second language (L2). Numerous studies on the positiveeffects of bilingualism on TLA relate the advantages evident in bilinguallearners to the influence of bilingualism on cognitive development and,specifically, to metalinguistic awareness (MLA) and communicative skills(Cummins, 1978; Thomas 1988; Cenoz & Genesee, 1998; Jessner, 1999;Jessner,2006;Jaensch,2009).Thishasbeenconsideredasanindirecteffectofbilingualism,that is, “bilingualismaffectscognition,MLAandcommunicativeskillsandthesefactors,inturn,affectTLA”(Cenoz,2003).

Although it is has beenwidely acknowledged thatMLA is strongly

38

affected by literacy and grammar related activities, only a few studies haveattendedtothecontextandmethodofacquisitionofthebilinguallearners’L2.Indeed, it is not clearwhether the higher level ofmetalinguistic awarenessresponsible for bilinguals’ better performance in TLA is mainly due to thepreviousinstructed,formallearningprocessintheirL2orifitistheresultofthe accumulated experience in a naturalistic bilingual setting. The currentresearch aims to investigate the bilingualism specific variables affecting theprocess and outcomes of TLA, with a particular focus on the relationshipbetween the context and method of acquisition of bilinguals’ L2 (i.e.instructed/uninstructed; formal/informal) and the degree of metalinguisticawareness developed. It is worth looking at how the level of explicit andimplicit metalinguistic awareness of “primary” and “secondary” bilinguals(Hoffman, 1991) correlates with a higher level of proficiency attained in athird(oradditional)language.Inotherwords,isitthelevelofbilingualismorthe level linguistic knowledge in an L2 that plays a fundamental role tosucceedinTLA?(Bialystok&Barac,2014)

Theproposedquestionwillbeassessed through thecomparisonofthegeneralproficiencyattainedinL3bybilingualswhoseL2hasbeenaddedin a naturalistic setting and bilinguals whose L2 is the result of a formallearning process. First, a bio-data questionnaire will report the learners’previous linguistic history, motivation, and aptitude towards languages aswellas theirsocio-economicbackground.Second, twocloze tests in theirL2andL3willbeadministeredinordertoassesstheirdegreeofbilingualism(i.e.balanced/unbalanced) and their level of proficiency attained in L3. Third, aSelf-PacedReadingTaskconsistingof twoparts, i.e. a linearnon-cumulativeprocessing of semantically ambiguous sentences and a post-experimentquestionnairefocusingonthegrammaticalacceptabilityofthesentences,willbe used in order to provide independent measures of implicit and explicitmetalinguistic skills developed in their L2 and information about theparticipants’ cognitive abilities (Blom & Unsworth, 2010) and anticipationskills(Bonifaccietal,2011).

SESSION11 WEDNESDAY31STMAY,11.25-12.40

11.25-11.50OnthestructureofpossessiveħagginfreestatenominalsinHailiArabicEisaAlrasheedi•NewcastleUniversity

PossessioninHailiArabic(HA)canbeanalyticallyformedbytheuseofthepossessivepreposition,ħagg,whichintervenesbetweenthepossessumandpossessor,asin(1).

(1)ʔal-bait ħagg ʔar-radʒaal (HA)

Def-house POSS Def-man‘Theman’shouse’The element ħagg is similar to the preposition li in the Modern

StandardArabic(MSA),comparetheexamplein(1)with(2),below.

10 39

here is the so-called double topicalisation construction, where two nominaltopics are found in clause-initial position.What kind of surface structure cancapturethevariousinternalorganisationsofdoubletopicalisationconstruction?

To answer this question, I take the Minimalist Program as a broadtheoreticalframework,andadopttheTopicPhrase(TP)analysis(Gasde&Paul,1996)whichtreatsTPasa functionalcategorywithmaximalprojectionaboveanIPorarecursiveTP,withnominaltopicsoccupyingthespecifierposition ofTP,andtheTopicheadcanbeeitheremptyoroccupiedbytopicmarkers.Inmyanalysis, I first show that in double topicalisation construction, the base-generatedtopicmustprecedethemovedtopic:(1a)[Hua], (meigui-hua)itazuixihuan_____iFlower rose-flower3SGmostlike‘Flowers,roseshelikesthemthebest.’(1b)*[Meigui-hua]i,(hua),tazuixihuan_____irose-flowerflower3SGmostlike*‘Roses,helikesflowersbest.’

This constraint on the relative order of base-generated topics andmovedtopicsmaybeexplainedbythe localityconstraintonmovement,whichfollows the idea of minimal effort that places shortness requirement onmovementoperations(Li,2000;Hornstein,Nunes,&Grohmann,2005).

Secondly,whenbothtopicsarederivedviamovement,theymustbeina‘nested’dependencyrelationwiththeirgapsinthecommentclause:

(2a)[zhejianshi]i,(lixiansheng)jwogaosuguo_____j_____iThis-CLmatterMrLiItellPERF‘IhavetoldMrLiaboutthismatter.’(2b)*[lixiansheng]i(zhejianshi)jwogaosuguo_____i_____jMrLiThis-CLmatterItellPERF‘IhavetoldMrLiaboutthismatter.’

ThismaybeexplainedbythePathContainmentConstraint(Pesetsky,1982),whichrequiresthepathsofmovementtobeinacontainmentrelation.

These findings challenge the view that Chinese topics are merelyconstrainedviaasemantic“aboutness”relationwiththecommentclause(cf.Xu& Langendoen, 1985). They suggest that topicalisation is subject to syntacticconstraintsthatgoverntherelativeorderofdoubletopicsinChinese,whichfallunder the broad notion ofminimal effort. Chinese has a rigidword order butlittle inflectionalmorphology,which sometimesmakes it difficult to pin downthegrammatical functionofaconstituentbasedontraditionaldiagnostic tests.Nevertheless, the kind of syntactic constraints described above allow Chinesespeakers to interpret double topicalisation constructions in a relativelyeffortless manner. The grammar that underlies Chinese topicalisation isdesignedassuch,sothatminimaleffortisrequiredfromthespeakertoresolvepotentialambiguitiesdespitethelackofovertmorphologicalcues.

9

First, the particles may co-occur on the same element, but two co-occurrencerestrictionscanbeobserved:theparticles–kOand–pA,ontheonehand, and – hAn and –s, on the other, cannot co-occur. Huhmarniemi (2012)argues that these restrictions arise from –kO and –pA expressing differentvaluesofthefeature[force],andsimilarly–hAnand–sencodingdifferentvaluesof the same feature. This suggests that the standardly postulated unitary CPmustbesplitfurthertoallowdifferentcategoriesofparticlestoco-occur.

Thispicturehastobeelaboratedfurtherbasedonthebehaviouroftheparticles in embedded contexts. I will adopt Haegeman’s (2006) typology ofembedded contexts, where different types of clauses are differentiated bywhether they encode Speaker Deixis (SD), a functional layer anchoring apropositiontoaspeaker.Whileperipheraladverbialclausesmirrorrootclausesin their structure in projecting SD, central adverbial clauses and clausesembeddedunderfactiveverbsdifferinmanifestingareducedstructurewithoutSD. Novel data show that the particles –hAn and -pA can occur in the formertypesofclauses,whiletheyareexcludedfromthelatter,reducedtype,providingevidenceforanSDlayerinFinnish.

Thedata show i) that theFinnish leftperiphery ismore refined thanstandardlyassumed,encodingaspeaker-andhearer-related layer,and ii) thatthe argument for discourse-related projections encoded in syntax iscrosslinguisticallycorroborated.

14.50-15.15ThesyntaxofsententialnegationinAbhaArabicdialectAliAlzayid•UniversityofEdinburgh

In this paper, I investigate sentential negation in the Abha Arabicdialect (henceforth AAD). It is a dialectal variation of Arabic spoken in thesouthwestofSaudiArabiaintheAsi:rregion.Tothebestofmyknowledge,Al-Azragi(1998)wasthefirstresearchertoexaminethesyntaxoftheAbhadialect.Itwasadescriptivestudyinnaturecoveringawidevarietyof topics includingnegation phenomenon. Based on her data, I argue thatAADhas four negationparticles: ‘maa’, ‘la’, ‘lim’, ‘lis’ as well as negative constructions in copularcontexts.Followingmanyworksadoptingwhatisknownas“Neg.Phypothesis”,I assume thatnegativeparticles inAADhead theirown functionalprojections.Moreover, the analyses of Benmamoun (2000) of Arabic functional categorieshavebeenadoptedwithslightmodificationstoaccountforthedatainquestion.15.15-15.40Doubletopicalisation,syntaxdoesconstrain!SherryYongChen•UniversityofOxford

Thispaperpresentsnewdatatoargueforsyntacticconstraintsontherelative order of nominal topics in double topicalisation constructions inMandarinChinese(henceforthChinese).

Chinese topics are derived via either base generation or movement(Badan&DelGobbo,2011;Huang,1982;Li,2000;Shyu,1995),andtheyalwaysappearinclause-initialposition.Beingatopic-prominentlanguage,Chineseisawindowintomanytypologicallyinterestingconstructions.Ofparticularinterest

(2)ʔal-bait-u li-r-radʒul-i (MSA)Def-house-NOM of-Def-man-Gen‘Theman’shouse’Saidthis,ħaggdiffersfromthelatterinthatitagreesinnumberand

genderwiththeprecedingDP,thepossessum(3a-b).(3)a.ʔas-sijjarah ħagg-at Mohammed (HA)Def-car.SG.F POSS-SG.F Mohammed‘Mohammed’scar’b.ʔas-sijjaraatħagg-aat Mohammed

Def-cars.PL.FPOSS-PL.F Mohammed ‘Mohammed’scars’

Whatseemsinterestingaboutħagg isthatithasshiftedfrombeinglexically used as a noun, meaning ‘right’, ‘righteousness’ or ‘truth’, to agrammaticalised functional item (a possessive preposition) denotingpossession.Moreover, thepresenceofanalyticgenitives, likeħagg, inArabicdialectshasbeenwidelytakenasa‘dialectalinnovation’(Harning1980).

Asforthesyntacticderivation,Iargue,followingDenDikken(2006),thatħaggcanbestructurallyderivedviaheadmovementofanullpreposition(Ro)headingtheRelatorPhrase(RP)toFooftheupperc-commandingLinkerPhrase(FP)formingacomplexhead,whereħaggisassumedtobeaPFformoftheamalgamatedhead(Ro+Fo).Thederivationrequiresfurthermovementto the Classifier Phrase (ClfP), situated immediately above the FP. Thismovement is necessitated by the involvement of the genitive preposition inpredicate inversion and in order to restore the original word order (DenDikken and Singhapreecha 2004). Additionally, the movement to classifierphrase is said to have an additional advantage of deriving preposition-possessum agreement by placing the genitive preposition in a Spec-headrelationwith thepossessum(DenDikken2006).Said this, Iargue further infavor of Ouhalla’s (2011) simplified version of the above in which theLinker+Classifier conflates into a single functional head, F[Agr] whichparallelsT[Agr]inclausesandcorrespondstoNuminDPdomain.ForOuhalla(2011), contra Den Dikken (2006), preposition-possessum agreement isassumedunderAgreerelationbetweenF[Agr]andthepossessumitcloselyc-commandsresultinginthevaluationofuφ-featuresofF[Agr]basedonthoseofthepossessum.Inthenextstepofthederivation,thePisraisedtoF[Agr]toget an inflected form of the preposition, i.e., ħagg, which agrees with thepossessum inφ-features. The final step comprises the possessum raising toSpec,F[Agr],yieldingtherightwordorder.11.50-12.15The noun phrase in central Kurdish: A projection of D (DP) not Num(NumP)RebwarTahir•NewcastleUniversity

This paper proposes that the noun phrase in Central Kurdish (CK,henceforth) is in factaprojectionof the functionalcategory(D), rather thannumber (Num). Although the surface form suggests that the functional

40

8

that-3SFstole-3SF DEF-carUngrammaticality of sentence (3) gives substance to the Kiparskian

stance that the structural difference between factive and non-factivecomplement lies in subcategorization of the matrix verb as far as JA isconcerned. With the assumption that complements of factive verbs are, withupdateterminology,DPswithasilentnounanddeterminer(Schueler2016),thetwofactsoffactivecomplementsarestraightforwardlyaccountedfor.WiththefactthatDPsareabsoluteislandsinArabicgrammar(Mohammad1999;Soltan2007), the observation that no extraction whatsoever is allowed from factivecomplementswhichareembeddedunderDP follows.Furthermore, aDP layerthatdominatesCPfactivecomplementshasnosyntacticeffectsonA’-movementwithintheseclauses,somethingthatgivesrisetothepossibilityoffronting(andevenCLLD)insidesuchcomplements.

Asforobligatoryuseofthecomplementizerʔinn,(theobservationthathas been considered a problem to the DP approach as the complementizer issaidtobearealizationtothepotentialoperator;cf.Zubizaretta2001)IappealheretoapromisinglineofresearchbySchueler(2016)whoassumesthat it isthecomplementizerbutnotthelexicalverbthatisresponsiblefortriggeringthepresuppositionsassociatedwithfactivepredicates.14.25-14.50Calling Finnish speakers: Evidence for speaker and hearer syntax fromdiscourseparticlesAnnaHollingsworth•UniversityofCambridge

Discourse particles are oftendismissed into the fringes of linguistics.However, recent work on discourse-related elements has revived the ideaattributed to Ross (1970) of representing the notions of speaker and hearersyntactically, based on imperatives (Alcázar and Saltarelli, 2014), temporalrelations (Giorgi, 2010), and evidentiality and logophoricity (Speas, 2004),amongotherphenomena.Thistalkcontributestothisgrowingbodyofevidencethrough theanalysisofFinnishdiscourseparticles.Myaimsare twofold: Iwillshow that discourse particles occupy an important role in syntactic theory ingeneral,andarguethatthecurrentviewoftheFinnishleftperipheryhastoberefinedtoincludeadiscourse-relatedlayer.

Iwill focusontheparticles-hAn(amarkerof familiar information), -kO (the obligatory interrogative marker) and –pA (marking emphasis orexclamativeforce),aswellas–s(amarkerofinformalregister).Theparticles–hAn, -pA, and –kO, but not –s, are second-position clitics, always triggeringmovementtotheedgeofC.

The Finnish left periphery has typically been assumed to be rathersparseintermsofprojections.Standardanalysespositspec,CPasthehostforabroad range of elements, including second-position clitics, wh-phrases, andcontrastivetopicsandfoci(Vilkuna,1995).Huhmarniemi(2012)elaboratesthispicturebyarguingfortwodifferenttypesofmovementtriggersinspec,CP:theobligatoryfeature[force]forphraseshostingparticlesandtheoptionalfeature[focus] for contrastive elements. However, considerations pertaining to co-occurrence restrictions of the particles and their appearance in embeddedcontexts show that amore refined left periphery is needed to account for thedata.

projection of number (NumP) is aboveDP in the structure, I argue that thenounphraseproperstillhasaDPprojectionaboveit.

Basedon typologicaldata fromawide rangeof languages,Rijkhoff(2002) claims that referentiality elements such as the definite marker (D)takesscopeover,henceisstructurallyhigherthanquantityelementssuchasnumber(Num),asin(1).

Likewise,thestandardstructuralrelationbetweenNumPandDPis

thattheformerfallswiththescopeofthelatter, i.e.DPishigherthan NumP(see Ritter 1991, among many others). Also, according to the currentgenerativeapproach,DPisstructurallyparalleltoCPwhereDcorrespondstoC,andNumPisparalleltoTP.Thus, if it istruethatCPiscross-linguisticallylocated above TP in the clausal domain, DP should be above NumP in thenominaldomain,nottheotherwayround.

On the faceof theseproposals, thehierarchical structure in theCKnounphraseseemstobeacounterexamplewhereDPfallswithinthescopeofNum. The number marking is realized by the enclitic -an attaching to thedefinitemarker–ekewhich,inturn,attachestothenominalprojection(NP).

(2)esp-eke-an

horse-DEF-PL‘thehorses’This study adopts Chomsky’s (1995) Minimalist bottom-up

derivational theory. Thus, observing the example (2), I take the nominalprojection(NP)esp ‘horse’tomergefirstwiththecategoryDrealizedbythedefinitearticle–eke(see3),beforemovingtoSpecDP.TheDPisthenmergedwithNumrealizedby–an,andmovesinapied-pipingfashiontoSpecNumP,asshownbelow.

Given this structure, the category D where -eke is realized is

structurallylowerthanthefunctionalprojectionofnumber(Num),suggestingthat the CK noun phrase is hierarchically represented as NumP, not as DP.This contentionmilitates against all the standardassumptions raisedabove.However,IarguethattheNumPin(3)stillfallsunderanotherDPprojectionwheretheDcategoryisnullinconstructionssuchas(2),butmorphologicallyrealizedbythedefinitemarker–ewhendemonstrativesarepresent.Iprovideempirical evidence that the morpheme –e, which co-occurs with

41

38

7

languageandCantoneseasadialect.Thestatusofthelanguagesreproduces,andallows speakers to perform, the political status of Hong Kong vis-à-vis thePeople’s Republic of China. Also consideredwill be Benveniste’s views on therole ofwriting in the ‘auto-semiotisation’ of language, and its ability to ‘short-circuit’ spoken language, aswell ashis insistence that society is, not reflected,butcontained,inlanguage.SESSION1 MONDAY29THMAY,14.00-15.4014.00-14.25AcloserlookatsyntaxofclausalcomplementsoffactiveverbsinJordanianArabicMarwanJarrah•NewcastleUniversity

Therearetwopossibleaccountsofimpossibilityof(subject)extractionout of clausal complements of factive verbs. On the one hand, factivecomplementshavebeenassumedtobeCPsheadedbyanullD°,somethingthatturnssuchclauses intostrong islands(cf.Kiparsky&Kiparsky1970,Manahlot1977,andOuhalla2005).Ontheother,theleftperipheryoffactivecomplementsobtains an operatorwhich blocksmovement out of these clauses (cf.Melvold1991, Hegarty 1992, Zubizaretta 2001, and Haegeman 2012). In this paper, Ipresent these twoaccounts andargue that theDPapproach is superior to theoperatorapproachinJordanianArabic(JA).

Firstly,accordingtotheoperatorapproach,A-barmovementtotheleftperipheryof factivecomplements isnotpermitted(Grimshaw1990,Makietal1999, and Zubizaretta 2001) or severely restricted; only Discourse-linkedelementscanbeextracted(Haegeman2012).Assuch,focalization,amongotherthings,wouldnotbepermittedwithinJAfactivecomplements,contrarytofact.Considerthefollowingsentence:

(1)ʔabuu-Iħizin/ʔistaʁrab/nasa/ʕirif ʔinn-haBINIT

father-myregretted/gotsurprisedof/forgot/knew that-3SF girlʔaχoo-isaragbrother-mystole.3SM

'Myfatherregretted/gotsurprisedof/forgot/knewthatitthatitwasagirlthatmybrotherhitatthemarket.’

Grammaticality of (1) is ample evidence that fronting to the leftperipheryoffactivecomplementsisnotdisallowedinJA.(1)ishardtoaccountforassumingthatthereisanoperator(intheleftperipheryoffactiveclausesinJA) blocking A’-movement instances to the left periphery, as all A-barmovementsarepermitted.Now,considerthefollowingill-formedexamplethatshows that no extraction (of arguments or adjuncts) is allowed to thematrixclause:

(2)*miinz-zalamih/weenʔabuu-I ħizin/ʔistaʁrab/nasa/ʕirifwhichman/whenfather-myregretted/gotsurprisedof/forgot/knewʔinn-hasarag-t ʔis-siyaarah

demonstratives(4),isthesyntacticrealizationofdefiniteness,servingasimilarfunctiontothemaindefinitemarker–ekein(2).

(4) ew esp-an-ethathorse-PL-DEF‘thosehorses’

Asexpected, thepluralnumbermarker–anoccurs inside thedefinite

marker, indicating that the D where –e is realized is above the category ofnumber(see5a).Comparing(5a)to(3)wherethenounphraseisaprojectionofnumber(NumP)withDPbelowit,IarguethatNumPin(3)stillhasasecondDPlayeraboveitwheretheDisnotrealized(see5b).

This unpronounced D category above NumP (5b) provides double

definiteness to the noun phrase, a phenomenon which is common in CK andseveralotherlanguagesincludingHebrew,NorwegianandSwedish(seeDanon2008;Julien2003).

12.15-12.50Variational models of language acquisition: The best is still not goodenoughCallumHackett•NewcastleUniversity

The Variational Model of language acquisition developed in Yang(2002) is a Principles & Parameters system which describes acquisition as aperiod of competition between all and only possible adult grammars. If weaccept that a grammar can be roughly characterised as the combination of auniversal base (UG), a suite of specified parameter settings, and a lexicon oflanguage-specific sound-meaning correspondences, grammar competition caneasilybecomeanintractablelearningtaskifitfollowsthatchildrenaremeanttoselectbetweenfully-specifiedparametersets(Roberts&Holmberg,2005).TheVariational Model’s solution to this is to recast grammar competition ascompetitionbetweenparameterenumerations.

Ratherthaneachpossibleadultgrammarbeingfullyrepresentedinachild’s mind, children have access only to some small number of individualparameters,eachofwhichhasanassociatedprobabilityofbeingenumerated[+]or [–].Each timea childhearsorproducesanutterance, they ‘hypothesise’ anadultgrammarbycompilingalistofvaluedparameters,thereafterrewardingorpunishingparameterprobabilitiesuntiltheyhaveallconvergedonavalueof0or1(thusensuringthatallparametersettingsinanadultarefixedas[+]or[–]).

Legate & Yang (2007) showed that Variational Models can berigorouslyappliedtocross-linguisticcorpusanalysesof‘RootInfinitives’(verbsthat children produce in non-finite form in contexts where finiteness isobligatoryintheirtargetlanguages).Thisabilitytoestablishadirectconnection

42

6

PLENARY1 MONDAY29THMAY,12.00-13.0012.00-13.00‘Languagecontainssociety’:Hong Kong’s Umbrella Movement interpreted through Benveniste’s LastLectures(andvice-versa)ProfJohnE.Joseph•UniversityofEdinburgh

Intheautumnof2014aprotestmovementwaslaunchedinHongKongagainst the government’s plan to constrain the electoral process for the ChiefExecutive.Througha complex chainof linguistic and semiotic associations theprotests quickly came to be called the Umbrella Movement, in Hong KongCantonese��ze1daa2“Chater”,literally“coverhit/beat”:ChaterRoadhavingbeenthelocationofearlyprotests,anditsfirstcharacter,�ze1,beingalsotheword for umbrella. Protesters had used their umbrellas to cover themselvesfrom the tear gas and baton attacks launched at them by police, hence thesignificance of the literalmeaning of ze1 daa2,which is opaque to speakers ofPutonghua.EvenmainlandCantonesecannotfullydecodethesignificanceofthemovement’sname.

Émile Benveniste (1902-1976) gave his last series of lectures in theCollègedeFrance in1968-1969.TheseDernières leçonsremainedunpublisheduntil 2012, and future historians might be forgiven for imagining that theUmbrellaMovementof2014wasdirectlyinspiredbythem.Itwasnot,ofcourse,andyetthemovement’ssemiologyprovidesevidenceforBenveniste’skeyideas.Benveniste laid out a semiological theory extending Peirce’s idea of the‘interpretant’ –whichmaintains that any sign ismeaningful only through theintermediaryofanothersign– to includethe languagesystemasconceivedbySaussure.ForBenveniste,othersignsystemsdependonlanguagetoactastheirinterpretant,andthesesystemsincludesocietyitself.

Theinterpretingsystemfurnishesthebaseoftherelationsthatpermit

whatisinterpretedtodevelopintoasystem.Thebaseisfurnishedbythelanguage: for example, the pronoun system, I/you versus he/she/it.Without this linguistic distinction that introduces the relationship ofdialogueandthatofotherness,nosocietyispossible.

Betweenthetwosystems,linguisticandsocial,thereisnostructuralcorrelation. The relation can only be semiological, namely a relation ofinterpretanttointerpreted,excludinganygeneticrelation.

Languagecontainssociety.Thelanguagecanbestudiedonitsown,asaformalsystem,without

taking account of society. The inverse is not the case. Society and therepresentations that govern it cannot be described outside linguisticrealisations.

Only what language denotes is social. (Benveniste 2012: 79, mytranslation)

How the umbrella took on the symbolism it did in Hong Kong in 2014 is asnebulous as any social creation of meaning. But deeply involved in it, asinterpretants, were the languages Putonghua and Cantonese – on the level oftheirwordsforumbrella(Putonghua�săn,HongKongCantonese�ze1),butalso on that of the whole existence of Cantonese and Putonghua as languagesystems, rather than as dialects of oneChinese language, or Putonghua as the

betweenachild’s input languageand theirproductionofRoot Infinitivesgivessuchmodelsmoreexplanatorypotential thanotherprominentaccountsof thephenomenon(e.g.,Rizzi,1994andWexler,1998).Iwillshowthatthesemodelscanberefinedevenfurtherbyusingmorenuancedsetsofparameters,sothatthesyntacticprocessesunderlyingRoot Infinitives,aswellas importantcross-linguisticdifferences,canbeevenmorerobustlycaptured.

Nonetheless,acoredifficultyfacedbyVariationalModels–andbythePrinciples&Parametersapproachtogenerative linguisticsasawhole– is thatputativeparametersthatarederivedfromcomparativedata(like ‘nullsubject’or ‘head position’) tend to encode descriptive facts in a technical system thatlooksexplanatorywithoutactuallyaddinganytheoreticalinsight.Thesourceofthisdifficultyissimplythatthereisaninfinitevarietyofbinaryparametersthat canbepositedtoaccountforcross-linguisticvariationandnounderlyingtheorytoconstrain them. Ina similarvein toBoeckx (2011), Iwill suggest that therecannot be any such constraining theory and that even the most attractiveparameter-based models of acquisition and adult grammars must necessarilyfallshort. SESSION12 WEDNESDAY31STMAY,13.30-14.45 13.30-13.55Acorpus-lexicographicaldiscourseanalysisoftheverb‘destroy’EmmaFranklin•LancasterUniversity

Thispaperpresentsaworkinganalysisoftheverb‘destroy’asusedinmodern British English, taking Hanks’ (2004, 2013) Corpus Pattern Analysis(CPA)approachasamethodologicalbasis.Todate,CPAhasonlybeenappliedincorpus lexicography and similarNatural LanguageProcessing tasks, but it hasthepotentialtoextendcurrentdiscourseanalysismethods.Assuch,thispaperdemonstrates an early attempt at a CPA-assisted, corpus-lexicographicaldiscourse analysis, using evidence from both a general corpus of English (theBritish National Corpus) and a specific, animal-themed corpus constructed byresearchers on the ‘People’, ‘Products’, ‘Pests’ and ‘Pets’ (PPPP) project (SealeyandPak, forthcoming).The study ispartof a largerproject–myPhD–whichseekstoestablishwhataCPAanalysiscanrevealaboutattitudestoanimalsbyexaminingtheirpositionswithin‘killing’verbconstructions.

CPA is underpinned by Hanks’ (2013) Theory of Norms andExploitations (TNE), a practical and empirical theory of language in the Neo-Firthian tradition, which posits that meaning is context-dependent and isinextricablylinkedtoform(cf.Sinclair(1991)).CPAinvolvesmanuallyanalysingthe co-texts of words (especially verbs), and from these concordancesidentifying syntagmatic patterns associated with senses of meaning. ThisprocessissimilartotheapproachtakenintheCOBUILDproject(Francisetal.,1996;Sinclair,1987),andsubsequentlyPatternGrammar(HunstonandFrancis,2000), but it places a greater emphasis on the semantic values of argumentsthanpreviouscorpus-drivenapproachestolanguagepatterns.Thismakesitanidealmethod for empiricallydemonstrating the typesof entities–or semantictypes – portrayed as participants in particular events. In the case of my PhD,these are the humans, animals and other entities that constitute ‘killers’ and

43

5

11.15-11.25 BREAK

SESSION11

11.25-11.50 Onthestructureofpossessiveħaggin

freestatenominalsinHailiArabic(p.39)

EisaAlrasheedi

11.50-12.15 ThenounphraseincentralKurdish:AprojectionofD(DP)notNum(NumP)(p.40)

RebwarTahir

12.15-12.40 Variationalmodelsoflanguageacquisition:Thebestisstillnotgoodenough(p.42)

CallumHackett

12.40-13.30 LUNCH

SESSION12

13.30-13.55 Acorpus-lexicographicaldiscourseanalysisoftheverb‘destroy’(p.43)

EmmaFranklin

13.55-14.20Master’sstudents’languageuseoftheirfirstandsecondlanguagesinclass:Asmall-scalecasestudy(p.44)

MingNi

14.20-14.45

StrategiesofpersuasionandargumentationinpoliticalrhetoriconexampleoftheinauguraladdressesofAmerican,British,Byelorussian,German,andRussian(p.45)

ElenaAfromeeva

14.45-15.00 BREAK PLENARY 15.00-16.00 Theacquisitionofcase-markingand

wordorderinheritagespeakers:GreekheritagechildreninNewYorkCity(p.46)

DrVickyChondrogianni

16.00-17.30 WINERECEPTION 18.30-19.30 PUBSANDWUGSPUBEVENING

-----

AlltalkswilltakeplaceinLG.09inDavidHumeTower(seemaponpage47).

BREAKSwillincluderefreshmentsoutsideLG.09.LUNCHwillbeinthe

commonroomonthe7thfloorofDugaldStewartBuilding.

Thereisliftaccesstoallfloors.

‘killees’incontemporaryBritishEnglish.Informedbytheobservationthattheverb‘destroy’isuseddifferently

for humans and animals, namely as a ‘killing’ termwhen applied to dogs andhorses,forexample,butnothumans,aCPAanalysiswascarriedoutonrandomsamples of concordance lines of ‘destroy’ from the BNC and the PPPP corpus.Thepreliminaryresultsindicatethattherearefoursensesoftheverb‘destroy’,asmarkedbytheirpatterns:twopertainingtodamageasinflicteduponphysicalobjects and abstract entities, respectively; one for referring to the defeat of ahuman group (e.g. in battle, sport, etc.); and one for denoting the killing ofunwanted or dangerous animals. The patternswere found to be stable acrossthetwocorpora,butthePPPPcorpussamplefeaturedmoreexamplesofanimal-relateddestruction (e.g. of animal homes rather thanhumanones) and raisedquestionsaboutthedistinctionbetweenhumansand‘nature’.Moreimportantly,it was found that by delimiting verb patterns using CPA, it is possible toempirically show when, and in what contexts, an entity is conceptualised asbelonging to another category. In this case, it was found that some animals –particularly wild animals – are subtly objectified by their roles in ‘destroy’constructions,asdemonstratedusingprecisedemarcationofverbpatternsandtheirarguments.

13.55-14.20Master’sstudents’languageuseoftheirfirstandsecondlanguagesinclass:Asmall-scalecasestudyMingNi•UniversityofSterling

Withthedevelopmentof internalizationandglobalization, there isanincreasingnumberofstudentschoosingtofurthertheirstudyabroad(Ilievaetal.,2004).Forinstance,theUKhoststhesecondlargestnumberofinternationalstudents in in the world (UK Council for International Student Affairs, 2015),with 438,010 international students during the academic year of 2015/16,accountingfor19.2%ofthetotalnumberofstudentsstudyingintheUK(HigherEducation Statistics Agency, 2017). And within these overseas students, thelargest proportion of them comes from China (Higher Education StatisticsAgency, 2016). For these Chinese students, English is the classroom languagebutnottheirfirstlanguage.Sohowistheiruseoflanguagesinclass?

In the presentation, I will report on an ongoing pilot for my PhDproject,whichlooksatChineseMaster’sstudents’languageuseoftheirfirstandsecond languages (i.e. Chinese and English) in class in the UK. Theseinternationalstudents’languageuseinsideclassroomhasdrawntomyattentionbecause I have noticed some interesting phenomena on some non-nativestudents’useoftheirfirstandsecondlanguagesinandafterclass(oneexamplecouldbetheytendtospeakChinesewithanotherChinese,evenifthereisanon-Chinesearound),andwouldliketofurtherinvestigateit.Theresearchquestionsofthisstudyinclude:

1. Underwhat circumstances do Chinese students use each language in

class?2. HowdoChinesestudentsexplaintheirchoiceoflanguage?3. Howtheprocessandmethodsofdatacollectionandmyparticipation

canbeimprovedformyPhDproject?

44

4

14.20-14.45 Iseewhatyoudidthere:Theroleoficonicityintheacquisitionofsigns(p.28) AshaStewart

14.45-14.55

BREAK

SESSION7

14.55-15.20 Freeadjunctsanddiscoursestructure(p.

29)JamesE.M.Reid

15.20-15.45 Constructiongrammarandlexicalpolysemy:Acasestudyoftheverbfeedintwoargumentstructureconstructions(p.30)

IvanaKralikova

15.45-15.55 BREAK

SESSION8

15.55-16.20 Ascientificrealiststancetowards

sociolinguisticmetatheory(p.31) JohannesWoschitz16.20-16.45 TheUnitedStatesof‘Merica:Indexicality

andTwitterhashtaguse(p.32) RachelMoyer16.45-17.10 ThechangingSwedish/iː/vowel:Evidence

fromthreecentralSwedishcities(p.33) FabienneWesterberg

17.30-19.30 POTLUCKPICNICINTHEMEADOWS 20.00-21.30 LANGUAGETOUROFEDINBURGH Day3:Wednesday31stMay

SESSION9

09.00-09.25 Linguisticnicheoracademicniche?Theroleoflanguage-externalbiasesinlanguageevolution(p.34)

JonasNölle

09:25-09.50 Theculturalco-evolutionoflanguageandmindreading(p.35)

MariekeWoensdregt,SimonKirby,ChrisCummins,&KennySmith

09.50-10.15 InvestigatingeffectsofmorphologicaltypesandtheirwordformationpatternsinSLA(p.36)

SvenjaWagner

10.15-10.25 BREAK

SESSION10

10.25-10.50 Thedevelopmentoftensemorphology

byWelsh-EnglishbilingualchildrenwithandwithoutSpecificLanguageImpairment(SLI)(p.37)

HyowonKwon&VickyChondrogianni

10.50-11.15 Instructedvs.uninstructedbilinguals:Theroleplayedbymetalinguisticawarenessinthirdlanguageacquisition(p.38)

FrancescaD’Angelo

Ethnographyhasitsroot inanthology(Fiona&Creese,2015). Inthisresearch,linguisticsethnographywillbeadopted,where Iobservedandparticipate inamoduleinaMaster’sprogrammeataUKuniversityfromintheSpringSemesteroftheacademicyear2016/17.FouroftheChinesestudentsinthismodulewerethe key participants, and what they said in class was audio recorded as theprimarysourceofdatainthisresearch.Andfollow-upinterviewswillbecarriedoutintheendwitheachkeyparticipanttofigureoutthereasonswhytheyuseChinese/English in class. The data collection for classroom conversation hasalreadyfinished,anditispossiblethatpartoftheinterviewscanbecarriedoutbeforetheendofMay;therefore,Iwillbeabletopresentsomeofthedatainthepresentation. 14.20-14.45Strategies of persuasion and argumentation in political rhetoric onexample of the inaugural addresses of American, British, Byelorussian,German,andRussianElenaAfromeeva•Friedrich-SchillerUniversityJena&BournemouthUniversity

The research pertains to Gerry Philipsen`s ethnographic study ofcultural communication (1981), especially on its` proposition that one of themajorfunctionofcommunicationistoconstructasenseofsharedidentity.Thus,it is postulated that every ethnic group possesses a distinct “code” which isdefinedas“historicallytransmitted,sociallyconstructedsystemofsymbolsandmeanings, primes and rules, pertaining to communicative conduct” (Philipsen1992:124,Philipsen2010).Thestudy investigates thedevelopmentofpoliticalrhetoric in the abovementioned countries in the light of their "code", i.e. inconnetion with their unique historical background and cultural peculiaritieswhichcreateanexclusivepolitical context.For instance, the researchassumesthat even identical rhetorical means will attain a wide range of connotationsdepending on cultural, historical andpolitical development of the society. Thestudyprovesthedependenceoflinguisticandstylisticdevices,whichembellishpresidentialoratory,onculturalandhistoricalcomponentsofthe“codes”whichtheypertainto.Theresearchinvestigatestheconnectionoflanguageandlinguo-stylistic devices which politicians use to manipulate public opinion andconstruct a sense of shared identity with cultural, historical, and politicalbackgroundtheyrelyon,inordertoapproachanidealimageofapolitician.Thestudy analyses variations of rhetorical patterns across cultures and theirreliance on a unique "code" of a specific community. The central idea of thestudy is rooted in the proposition that in order to be more persuasive andeloquent,presidentsappeal to the featuresof their “codes”andapply identicallinguisticandstylisticmeansindifferentconnotations.

Central are the notions of persuasion and argumentation in politicalcommunication with special focus on the inaugural addresses delivered byAmerican,British,Byelorussian,German,andRussianpoliticalleaders.

The objectives of the study can be practically demonstrated in thecorpus of the political speeches which is aimed to be created for the publicusage. The practical analysis ismade bymeans of GraphAnno programme for

45

3

18.05-18.30 TheinfluenceoforalreadingonKoreanlanguageproficiency:Acasestudy(p.14)

JihyeEo

18.30 CONFERENCEDINNER Day2:Tuesday30thMay

SESSION4

09.00-09.25 Scarystatisticsandhistoricallinguistics(p.15) DaisySmith

09.25-9.50 Language,identityandcommunity:Theuseofthird-personneuterpossessivesofearlymodernpreachers(p.16)

HiroshiYadomi

9.50-10.15 DiscourseandtopicalizationinoldEnglishsubordination(p.17) SergioLopezMartınez

10.15-10.25 BREAK

SESSION5

10.25-10.50 Mechanismsofcognitivecontrolinbilinguals:Theempiricalgeneralizabilityofthe‘bilingualadvantage’(p.17)

MichelaBonfieni

10.50-11.15 EffectsofconceptualaccessibilityandsimilarityinsimplesentenceproductioninMandarin(p.18)

YangziZhou,HollyBranigan,&MartinPickering

11.15-11.40 SyntacticprimingandanimacyeffectsinRussian-speakingchildren(p.19) AlinaKonradt

11.40-12.05 Parsingwh-questions:EvidencefromL1-Greekadultsandimplicationsforbilinguallearning(p.20)

KaterinaPantoula

12.05-13.05 LUNCH&POSTERSESSION

Theinfluenceoftemporalcontextonthe

productionoftemporalmorphologyinL2speakersofEnglish(p.21)

QingyuanGardner,VickyChondrogianni,&HollyBranigan

Indirectnessinthefaceofculturalcompetency(p.23) MayOuma

Phoneticpreaspirationofword-finalvoicelessfricativesinNorthEastScotland(p.23)

DavidA.J.Warren

Exploringyounglearners’foreignlanguageanxietyinChina(p.24) XiaoyiHu

SESSION6

13.05-13.30 Iteratedlearningoptimizesforsimplicity

(p.25) JonCarr

13.30-13.55 Populationdynamicseffectsontheevolutionofcommunicativeconventions(p.26)

JoséSegoviaMartín&MónicaTamariz

13.55-14.20 Squaringaline:Howsimilararegradableadjectivestoquantifiers?(p.27) FaustoCarcassi

creating graph-based annotations, where all meaningful parts of speech aredividedintocategories,forinstance,“state”,“religion”,“values”,etc.Theanalysisis made in the context of the “codes” of the given countries (in the light ofhistorical,cultural,andpoliticalexperienceofthesociety).PLENARY2 WEDNESDAY31STMAY,15.00-16.0015.00-16.00The acquisition of case-marking and word order in heritage speakers:GreekheritagechildreninNewYorkCityVickyChondrogianni•UniversityofEdinburgh

Heritage speakers are bilingual speakers who acquire the minorityheritage language frombirthathomeand themajority languageof thesocietythrough schooling and the wider community (e.g. Montrul, 2016; Polinsky &Kagan, 2007). Heritage speakers from immigrant communities, in particular,typicallyhaveweakerabilitiesintheirheritagelanguageandhavebeenshownto perform differently from monolingual speakers of the same language whogrowupinthecountryoforigin.Areasof languagethatareusuallyvulnerableare those where crosslinguistic differences between the minority and themajority languages occur.Much research on heritage speakers has focused onadults with fewer studies examining the factors that affect heritage bilingualchildren’sdevelopmentoftheirheritagelanguage.

In this talk, I present a series of production, online and offlinecomprehension experiments examining the linguistic (case ambiguity, wordorder) and extralinguistic (heritage language proficiency and use) factors thataffectheritagechildren’sabilitytointegratewordorderandcase-markingcuesin sentence production and comprehension when the dominant language(English)offersevidenceforwordordercuesbuttheheritagelanguage(Greek)utilises case-marking cues or conflicting word order cues. The results arediscussedwithincurrenttheoriesofheritagelanguageacquisitionandbilingualdevelopment.

46