Vegard Nygaard Bergh A study of young Norwegians in English Idiom Processing

Transcription

1 Vegard Nygaard Bergh A study f yung Nrwegians in English Idim Prcessing Master thesis in English linguistics Nrwegian University f Science and Technlgy Faculty f Humanities Department f Language and Literature

2

3 Abstract A quantitative study was cnducted with a grup f twenty-six Nwegian upper secndary students and a grup f twenty-six Nrwegian university students. Bth grups were tested fr English prficiency and idim cmprehensin. A third grup f English native speakers was als given the idim tests, t judge whether the tasks and idims were apprpriate. The idim cmprehensin tests were divided int tw part. In ne they wuld respnd t written alternatives, in the ther they wuld respnd t image alternatives. The upper secndary grup was measured fr their muse-mvements with MuseTracker sftware when perfrming the image idim test. The study aimed t cmpare the tw grups, and analyse the prcessing f idims particularly fr the upper secndary grup. The results shwed that the university students utperfrmed the upper secndary significantly in bth the prficiency tests and the idim tests. The university students als utperfrmed the English native speakers. The results f the muse-mvement tracking shwed that the upper secndary students prcessed varius idims in different ways, and that they were ften uncertain in the prcess f selecting a crrect respnse. The reasn why sme idims prved mre prblematic than thers was discussed, but n definite reasn was cncluded. iii

4

5 Acknwledgements The wrk put int this thesis has been as challenging and, at times, frustrating as it has been rewarding. The prcess f creating the entire wrk that makes this thesis has given me an experience f fulfilment. I had never imagined in my earlier life that I wuld ever cmplete smething s demanding and s challenging. I wuld like t thank my supervisrs. Mila Vulchanva fr sharing her knwledge and ideas, and supprting me alng the rad. Gisuè Baggi fr helping me a great deal with statistics and everything else, really. A big thanks shuld als be directed t Hendrick Rik Ertshuis wh aided me with his incredible skills with SPSS and the MuseTracker sftware. Even thugh he was nt appinted as a supervisrs, he put in many hurs t help me create smething ut f raw data. I als want t thank Anders Hauge Aurland fr his camaraderie as my prject partner. He spent many hurs n wrking with varius tasks, especially his wrk with the drawings that represent ur image idims in this thesis. I als want t direct a special thanks here t my grandmther, Gunvr Andb Nygaard. She is a prfessinal writer wh published numerus wrks in Nrwegian, frm fantasy nvels t shrt stries. Her inspiratinal being and immensely creative fantasy has inspired me thrugh my entire life, nt nly t achieve smething big, but als as a creative persn. I thank her fr ur talks and discussins, fr being patient, fr being as easy t agree with as t disagree with. I want t thank my friends and clleagues frm the English teacher s training prgramme.. A last thanks is directed t my best friend, Malin Berge, wh supprted me thrugh these years. Her daughter f 6 mnths, Sfie, was als a light that shined my wrking days with the thesis brighter. Bergh, Vegard Nygaard Trndheim, May v

6 Preface Relevance fr my future jb as a teacher This thesis is relevant t my teaching career mainly in tw ways. The first reasn is based n the experience f having researched smething and analysed the results. Thrughut the five years n this prgram, the university has cnsistently urged us as teacher t bring ur research t the class. In pedaggics, researching in the classrm is a ht tpic. Teachers can use this in practice mre than anyne else. They meet the same students cnsistently, and are the nes t see the students prgress best. T research yur practice in teaching is t adjust t the students, and thus yu can imprve yur teaching fr the next generatins f students. The secnd reasn is grunded in my knwledge and study f the English language and Secnd Language Acquisitin. The Nrwegian schl takes teaching f English very seriusly, and in an ever glbalizing wrld with English language as a key cnnecting factr, I see this as mre imprtant than ever. Having researched hw idims are prcessed, I investigated and scrutinized the fields f Secnd Language Acquisitin, Crss-linguistic influence, Metalinguistic awareness and abilities, and much mre. This study sparked my curisity twards the entire field, and made me aware f the imprtance f utside factrs influence n L2 acquisitin. Metalinguistic awareness, a mix-up f knwledge frm all kinds f different fields and nt language itself, is incredibly interesting. That a student can read a figurative expressin never befre seen, with wrds and phrase-structure that makes n sense, but still be able t make sense f it is remarkable. I will take the infrmatin I gained frm this study very seriusly, and apply methds f teaching that can imprve metalinguistic awareness in the students, as a cmplementing methd t my general English teaching. vi

7 Table f Cntents 1. Intrductin... viii 2. Theretical Framewrk Representatin and Prcessing f idims The mental lexicn What is stred and what is cmputed n-line? Cmpsitinal and Nn-cmpsitinal appraches The influence f familiarity n prcessing and representatin Age, develpment and cntext Idims in Secnd Language Acquisitin Metalinguistic awareness Methds and Prcedures Participants The native speakers Materials and Prcedures The English Prficiency tests The Idim tests Analysis The English Prficiency tests The idim tests Ptential Surces f Errr and Limitatins Results The English Prficiency Tests Idim tests: Accuracy Idim test: Reactin Times Discussin The English Native speakers The English Prficiency test and Idim Accuracy MuseTracker data Reactin Times Respnse Certainty General Discussin Cnclusin Questins fr further research vii

8 References Appendices Appendix 1: Cnsent Frm Appendix 2: Backgrund Questinnaire Appendix 3: Idim tasks with written cntext Appendix 4: Idim tasks with image alternatives Tables and Figures Figure 1: Idim task with image alternatives...21 Figure 2: Example f raw MuseTracker trajectry data...25 Figure 3: Example f remapped MuseTracker trajectry data 25 Figure 4-5: Bxplt f Crrect respnses and incrrect respnses fr bth grups by bth stimuli...29 Figure 6: Reactin Times per Item.32 Figure 7: Maximum Deviatin in crrect respnses.33 Figure 8: Maximum Deviatin in incrrect respnses..33 Figure 9: Area Under the Curve in crrect respnses..34 Figure 10: Area Under the Curve in incrrect respnses Figure 11: Fur selected idim trajectries...44 Table 1: Grammar and Vcabulary tests by Mean and Standard Deviatin 27 Table 2: Idim respnse accuracy..28 Table 3: ANOVA..30 Table 4: Reactin Times by Mean and Standard Deviatin 31 viii

9 1. Intrductin Idims represent a bulk f everyday language, and they represent a grup f figurative expressins which benefit frm a knwledge that exceeds linguistic skills. Idims can cme in very many sizes and frms. Sme f them are s transparent in cmpsitin, that ne can read them and understand frm the wrds what the meaning is. Other idims seem t be cmpletely arbitrary, carrying cultural etymlgies in them with meanings frm a bygne age. These idims can be clse t impssible t derive a meaning frm unless yu have acquired a certain knwledge. But what is this knwledge exactly? What divides a gd idim cmprehender frm a bad ne? These questins are recurring thrughut this thesis, and will be addressed alng with many ther factrs fr idim prcessing. The Nrwegian Directrate fr Educatin and Training emphasizes the additin f idim teaching in their English subject curriculum. Hwever, the directrate allws fr each and every teacher t carry the respnsibility fr what they chse t teach and hw. Nrmally they allw fr everything t be part f the classes, but nt always d all items frm these lists get the same attentin. Knwledge f idimatic expressins culd bst the cmprehenders metalinguistic awareness, which in turn culd grant access t understanding a series f ther types f expressins with interlingual features. This is a crucial part f acquiring a secnd language. Nrwegians are usually knwn t be relatively skilled in English as an L2. Des this mean that they are equally gd with metalinguistic skills? D they perfrm well in English prficiency as well as with idims? Such questins sparked the curisity f the experimenters perfrming this study, and hpefully we culd grant a small insight int the tpic. In the fllwing, I will present a theretical framewrk f the nature f idims, a backgrund fr the field and the mst imprtant studies and findings surrunding the prcessing f idims. Next, I will describe the methds and prcedures as a fundatin fr this study, befre I present the results with descriptive statistics. Lastly, a discussin f the findings with pssible explanatins fr the results will be presented befre ging t a cnclusin. The literature presented in the theretical framewrk will be used t back up the discussin f the results and the methds used. The executin f the study was carefully planned, and all the tests were cnstructed by us thrughly. 1

10

11 2. Theretical Framewrk 2.1 Representatin and Prcessing f idims The nature f idims Althugh it is widely accepted that idims must be stred in the lng-term memry and that they are stred similar t wrd-like lexical items, it is discussed hw they are prcessed differently frm nrmal cnstructins. They have ften been classified as figurative r nnliteral language tgether with expressins like irny, clichés, metnymy, sarcasm and ther cnstructins that require the cmprehender t lk beynd the literal meaning f the utterance and lk fr a meaning (Titne & Cnnine, 1999). Levrat & Cacciari (2002) argued that expressins f figurative nature are very frequent in daily language and Jackendff (2002) suggested that idims alne amunt t thusands, pssibly in prximity t the amunt f adjectives, despite a tendency amng traditinal grammarians t marginalize idim frequency. Pll, Barlw, Fine and Plli (1977) analysed plitical debates, nvels, psychlgy texts and mre, and fund that mst English speakers utter arund 10 millin nvel metaphrs and 20 millin idims in their lifetime. Idims are defined in the Cllins Cbuild Dictinary f Idims (1995) as a special kind f phrase. It is a grup f wrds which have a different meaning when used tgether frm the ne it wuld have if the meaning f each wrd were taken individually (p. v). This reflects the duble nature f idims: On ne hand they are part f the lexicn as large items and are stred similar t ther lexical items like wrds and mrphemes. On the ther hand they are cmplex structures and may be prcessed as grammatical structures abve the level f lexical items (Vulchanva, Vulchanv, & Stankva, 2011). This is what makes cmprehensin f idims fr children difficult, as well as analysing and classifying them is a prblem fr linguists. The idims mstly cme in a pre-packaged frm where parts usually cannt be replaced r mved. Jackendff (2002), discusses the flexibility f varius idims well in his paper What s in the Lexicn?. The famus idim kick the bucket, fr example, has a syntax which seems t be cmpletely deviant frm meaning. It cannt be passivized nr its cnstituents remved r mdified in any way and still hld the idimatic meaning. He als lists sme idims that can, in varius degrees, be subjected t syntactic analysis, like give NP the bt and V NP s head ff. By this reasning the idims mve acrss degrees f fssilizatin, frzenness and semantic transparency (Vulchanva et al., 2011). A typlgy f idims, then, will have t accunt fr the fixed/nn-transparent and flexible/semi-transparent expressins as suggested in 3

12 Jackendff (2002). Vulchanva et al. pint t an interesting trend here, that it shuld be kept in mind that there is an interesting crrelatin between the degree f flexibility in frm and semantic interpretatin the mre fixed the surface frm f the expressin is (i.e. the mre frzen), the mre de-semanticised its cmpnent parts becme. (p.143). It is necessary t have a mdel t predict hw idims are represented and prcessed, and in fact there are several ways f understanding idims fr this purpse. The theries n hw idims are represented and prcessed cnverge n tw main appraches; namely the Nncmpsitinal and the Cmpsitinal theries The mental lexicn This tpic is deeply embedded int the field f psychlinguistics as a rt frm the idea f the language faculty as a bilgical result f evlutin. The scpe f the bilgical accunt is based n the idea that we minimize the amunt f energy expended fr the mst efficient cmmunicatin prcess. We knw much abut hw wrds are structured and what wrds and their respective mrphemes cmpse in terms f meaning. The questins are hw the brain acquires rules f hw the prefix un- results in a meaning that ppses the riginal wrd, where and hw knwledge is stred r hw it decides t break wrd cmpnents dwn (Altmann, 1997). Frm the 1970 s research n hw humans perceive and cmprehend language grew. Fr instance, Marslen-Wilsn demnstrated that we can recgnize wrds even befre they are fully heard and that the reactin times fr hearing them was arund 200 millisecnds (Marslen-Wilsn, 1987). They fund that wrds wuld be distinguishable nly when the listener reached a certain pint f the utterance. S when a wrd becmes distinguishable, it wuld have reached its uniqueness pint (Altmann, 1997). In fllwing sectins n prcessing and representatin f idims, a hypthesis that suggests such a mental trigger pint als fr the cmprehensin f idims, namely the Idimatic key will be presented. The mental lexicn discussin in Jackendff s paper (2002) builds n the research f rdinary linguistics ging back t Chmsky s Aspects, in which Chmsky lays the fundatin fr the structure f grammar as cntaining a phnlgical, semantic and a syntactic cmpnent. The syntactic base generates deep structures, which accunts fr grammatical rules and phrasestructures, and tgether with lexical items like wrds and mrphlgical units they map with the semantic and phnlgical cmpnents t be cmprehended by the human brain (Chmsky, 1965). His generative grammar was explained with a basis in evlutin as a 4

13 cgnitive capacity, and set a standard fr hw linguistic thery was suppsed t prceed. Yet methds and years f research has taken the field much further. It was in Aspects he intrduced the wrd lexicn as a strage fr all the wrds the speaker knws. He argued further that the lexicn must cntain all exceptins and variants f the wrds, and that there exists n redundancy in terms f lexical variatins. The mst relevant part f his thery, als the mst relevant fr this theretical framewrk, is Chmsky s treatment f the creative r generative pwer f language as part f the syntactic cmpnent that cmbines lexical items int phrase-structure. In his article, Jackendff (2002) further explains Chmsky s treatment best: By cntrast, the phnlgical and semantic cmpnents f language is taken t arise by means f a derivatin frm syntactic structure. Mrever, lexical items (wrds) are taken t be inserted in their entirety int syntactic structures; their phnlgical and semantic aspects are carried ver int phnlgical and semantic structures respectively thrugh the prcess f interpretatin. (p. 25). Jackendff argues fr a different view explaining the lexicn, shwing evidence fr issues with the grammarian traditin, interestingly by shwing the cmplicatin f idims, and prpses his wn apprach in the fllwing sectin What is stred and what is cmputed n-line? Jackendff argues that evidence cming after Aspects shws that phnlgy and semantics bth seem t be incmpatible with Chmsky s lexical representatin (Jackendff, 2002). His cnditins fr questining the traditinal mdel were the questins 1) What aspects f a spken utterance must be stred in lng-term memry, and 2) what aspects can be cnstructed n-line. He ges n t argue a redefinitin f the lexical items, suggesting that a lexical item can be very different frm transfrmed wrds. By cmparing Turkish language, fr instance, where wrd stems are cmbined with a series f inflectins, with English, he argues that the variatins a wrd can have with mrphemes attached lgically shuld be prcessed by a different mechanism than the syntactic lexical ne. It simply seems very inefficient fr the brain t stre all pssible cmbinatins f mrphemes and wrds in the lng-term memry (Jackendff, 2002). The mst interesting part f his arguments in that paper, and fr my thesis, is the prblem f idims fr traditinal grammar. He criticised Chmsky and thers fr treating idims as lexical categries s that they can fall in under syntactic trees like everything else. The mst famus idim example thrughut linguistic research, and 5

14 repeatedly mentined in papers n figurative language is the nn-cmpsitinal idim Kick the bucket. This, as well as ther examples f idims f varius degrees f cmpsitinality, Jackendff presents as cunter-evidence t the grammarian view n lexical items and strage. He argues further that this has cnsequences fr the theries f language acquisitin, that besides the prblem f learning wrd meanings with the inherent grammatical structure knwledge, there is reasn t believe that the lexical knwledge is frmed t unify pieces frm a cmmn frmat as a rule. He ges n t say that: The meanings f cnstructinal idims are abut as cmplex as the meanings f wrds. Cnsider fr instance the time-away cnstructin, which means rughly spend time wastefully r frivlusly ding smething. This meaning is cmparable in cmplexity and subtlety t, say, prcrastinate, spend time wastefully in rder t avid ding smething, and s it shuld nt be mre difficult t learn (p. 53). The imprtance f a gd apprach t idims in linguistics is crucial, and in the fllwing sectin I will present the varius hyptheses and cnditins fr idim cmprehensin Cmpsitinal and Nn-cmpsitinal appraches The tendency fr nn-cmpsitinal theries is that idim cmprehensin mainly cmes frm memry retrieval, but they differ cncerning hw the idim meaning is accessed (Titne & Cnnine, 1999). Bbrw and Bell (1973) prpsed a literal prcessing mdel, knwn as the Literal First Hypthesis, in which a mental list f idimatic expressins is accessed by a special idimatic mde f prcessing. Upn reading an idimatic expressin, the cmprehender first attempts a literal interpretatin, and if this fails, the figurative meaning activates by this idimatic list. The methd f research used t back the Literal First Hypthesis, hwever, was criticised by Swinney and Cutler (1979) fr relying n self-reprt measures f cmprehensin. Swinney and Cutler prpsed their wn mdel, the Lexical representatin mdel, r Simultaneus Prcessing Hypthesis, which was based n idimatic expressins being stred and retrieved similar t lng wrds. The mdel suggested that a figurative meaning wuld fllw frm cmputing already the initial wrd f an idimatic expressin. This prcess invlves tw parallel prcesses: the retrieval prcess and the literal cmpsitinal cmputatin. The literal cmputatin activates immediately upn reading a string, and the cmprehender attempts t cmpute a meaning frm the cnstituents, but as this takes lnger in the case f idims, the retrieval prcess wins because the meaning is already stred in the lng-term memry in the mental lexicn (Vulchanva et al., 2011). The Direct 6

15 Access mdel, prpsed by Gibbs (1980), takes the figurative prcessing a step further and hlds that nly an idimatic meaning f the phrase is available during the cmprehensin. Thus it gives the idimatic meaning pririty ver the literal cmputatin, based n the argument again that the figurative meaning is retrieved much faster than the literal can be cmputed. The cmmn grund fr these nn-cmpsitinal hyptheses are that they share the assumptin that the idimatic meanings f certain cnfiguratins are semantically different frm the literal meaning f the cmpnents (Titne & Cnnine, 1999). These hyptheses d hwever have sme issues. They are cmmnly criticised n the scpe f their typlgy, and their failure t accunt fr idim flexibility. They apply well t frzen, nntransparent idims such as kick the bucket, but fail t encmpass flexible idims like spill the beans and transparent idims whse literal meaning smetimes may be mre easily interpreted than the figurative meaning. As explained frm the reasning f Jackendff (2002) n the string f idimatic expressins, it is prblematic t assume that idims with mdificatins can be stred fr all their variatins, n the accunt f efficiency. Vega-Mren (2001) suggests that it is prblematic t assume that frzen idims can be literally paraphrased. She argues that t view kick the bucket as meaning the same as the verb die has prblems. It is nt true that kick the bucket wuld be used fr all situatins where die wuld. Fr example, ne wuld nt use the idim in a phrase like A man kicked the bucket in a car accident tday. Cacciari and Tabssi (1988) cnducted experiments with crss-mdal priming, using a mre accurate methdlgy fr testing n-line meaning activatin. They cncluded that the final wrd f the idim shwed high predictability and determined activatin f cmprehensin. It was als fund that the idim meaning cmprehensin ccurred nt at the ffset, but 300 millisecnds after. They thus suggested that the idim meaning is a distributed representatin rather than a lexical entry. Later this was termed the Cnfiguratin hypthesis, claiming that the idim meaning is retrieved nly upn hearing a certain amunt f the string. This pint was called the Idimatic Key (Cacciari & Glucksberg, 1991). Building n this, Glucksberg later prpsed the Phrase-Induced Plysemy (PIP) hypthesis f idim cmprehensin, suggesting that idim strings are prcessed literally until a cnfiguratin realizes sme time after the ffset and an idimatic meaning is activated (Glucksberg, Brwn, & McGlne, 1993). This, tgether with the Cnfiguratin hypthesis frm ne f the mst cmpelling accunts f idim prcessing. 7

16 The cmpsitinal apprach accunts fr the cmpsitinality idims can have. The idea is that the literal wrd meanings and the cntext play a rle in representatin and prcessing f an idim string. These idims have been categrized int three grups by Geffrey Nunberg (1978), which he described as a cntinuum f cmpsitinality. The nrmally decmpsable idims in which a part is used literally, like fr example pp in pp the questin r save in save yur skin. The abnrmally decmpsable idims in which a referent f an idim s parts can be interpreted metaphrically, like maker in meet yur maker. The third categry is the semantically nndecmpsable idims which fits the traditinal view because the meaning cannt be cmpsitinally derived frm the wrds f the string, like chew the fat. Interestingly, Gibbs et al. (1989) wh researched idim prcessing based n these three categries, fund that subjects read sentences cntaining decmpsable idims faster than sentences with nndecmpsable idims, indicating that a cmpsitinal analysis is initially attempted. The cmpsitinal apprach seems t treat idim representatin and prcessing much like literal language. It acknwledges and accunts fr semantic and syntactic flexible idims, and assumes that we interpret idims literally as well as figuratively during cmprehensin. Hwever, the fact that the cmpnent wrds aid the cmprehender in interpreting idims des nt rule ut the pssibility f a pre-packaged meaning fr certain cnfiguratins f wrds (Titne & Cnnine, 1999). Furthermre, the fact that idimatic sentences are prcessed faster than nn-idimatic sentences, and the manner in which familiarity plays a rle in the prcess must be accunted fr (R. Gibbs, 1980; Swinney & Cutler, 1979). It is clear that the cmprehender s familiarity twards the given idim plays an imprtant rle, ne that will be explained in the fllwing The influence f familiarity n prcessing and representatin The majr idim prcessing theries have been t rigid fr gd methdlgy and preceding research has been incnsistent in testing acrss the varius idims, accrding t Titne and Cnnine (1994). They discuss fur dimensins alng which idims can vary and their relevance t idim prcessing. One f them is cmpsitinality and the thers are predictability, literality and familiarity. Familiarity in this cntext can be defined as the frequency in which cmprehenders encunter a given idim, and the degree t which it is knwn r easily understd (Gernsbacher & Glucksberg, 1984). Gernsbacher and Glucksberg fund that familiarity influences wrd recgnitin. It is, in few wrds, ne f the factrs fr which idim cmprehensin tests shuld be cntrlled. In her paper Representing and Prcessing idims (2001), Vega-Mren makes a great accunt n prcessing and 8

17 representatin f idimatic expressins. Her arguments divide prcessing fr whether the idim is familiar r unfamiliar. She suggests a relevance-theretic accunt f idims, prpsing a hypthesis: that idims are mentally represented and prcessed as structured phrasal cncepts and understd fllwing cnsideratins f relevance. As the idim is heard, bth the cncepts underlying the individual cnstituents in the string and the cncept underlying the idim as a hlistic unit are activated. Precisely which f this activated infrmatin is accessed and prcessed n-line fllws frm cnsideratins f relevance. On the ne hand, idims are seen as cnceptual units. They encde cnceptual representatins that have n equivalent in any nn-idimatic linguistic string and thus cannt be paraphrased withut lss. On the ther hand, since the cncepts underlying idims have internal structure, they are prcessed in much the same way as ther nn-idimatic sequences f wrds. (p. 83). She hlds that althugh the wrd meaning in idim cmprehensin des play a rle, it might be accessed n-line, highly cnstrained by pragmatic cnsideratins and nly at certain pints f the prcessing. On acquiring unfamiliar idims, she stated that the prcessing is a blend between sme f the encded material in the string used tgether with backgrund knwledge as a starting pint fr pragmatic inference. Her discussin f hw unfamiliar, cmpletely paque idims are acquired is mst interesting. In this case the cmpsitinal reading shuld be f n help in reaching a meaning, and current cmpsitinal appraches give n explanatin fr this (Vega-Mren, 2001). She argues that the nly pssibility fr the cmprehender is t access assumptins and implicatins based n the immediate cntext and backgrund knwledge. If there is insufficient cntext, hwever, the cmprehender will likely nt be able t create the meaning n-line. The case fr unfamiliar idims is thus a relatin between tw cnceptual representatins, the cmputatinal prcessing and the intended idimatic meaning. Fr familiar idims the string is decded and retrieved as an idimatic meaning withut having t cmpute the entire string literally. She suggests that the familiar idims are stred as cmplex structured cncepts cnsisting f three entries. A lgical entry which links with ther lgic cncepts, an encyclpaedic entry which wrks as a srt f schema with internal structure, and a linguistic entry which cnverges much with the abvementined cnstructinal idim representatin in Jackendff (2002). This framewrk agrees with the tendency twards a synthesis thery f nn-cmpsitinal and cmpsitinal appraches. Idimatic expressins and especially highly familiar idims are prcessed faster than nn-idimatic nes. Whether r nt an idim s cmpnent wrds 9

18 explicitly cntribute t idimatic meaning, the idimatic expressins are highly verlearned cnstructins f wrds cnsistently tied with specific phrasal meanings (Titne & Cnnine, 1999). As fr the cmpsitinal part, we knw that wrd meanings are always activated during idim prcessing (Cacciari & Tabssi, 1988), and that the wrd cmpnents substantially cntributes t the n-line cnstructin f idimatic meanings. Thus it is believed that bth the literal analysis f the phrase ccurs simultaneusly with the figurative retrieval which ccurs upn encuntering the idimatic key (Titne & Cnnine, 1999). The human ability t create ad hc cncepts nearly instantly allws fr cmmunicatin f un-encded meanings withut truble, and this is what Vega-Mren (2001) claims in her Relevance Thery, that ur pragmatic inferential abilities can decrypt the encded cnceptual material that makes idimatic expressins. These theries f prcessing and representing idimatic expressins must be cmplemented by a factr f age, which has been cvered by much research. Differences in idim cmprehensin acrss age will be the tpic fr the fllwing sectin. 2.2 Age, develpment and cntext Studies f idim cmprehensin and acquisitin have underlined three strng factrs t determine the ease f idim understanding: the familiarity f the idim string (Levrat & Cacciari, 1992), the semantic analysability an idim s cnstituents and the cntext n which the idim is encuntered (Levrat & Cacciari, 1995). Nte that these factrs have been tuched in the theries n prcessing and representing f idimatic expressins already, fr instance by Jackendff (2002) and Vega-Mren (2001). An interesting mdel fr the acquisitin f figurative language was prpsed by Levrat and Cacciari (1995) called the Glbal Elabratin Mdel (hencefrth GEM). The GEM emphasizes that n special prcedure r pre-knwledge shuld be presuppsed with respect t hw children nrmally prcess language. Particularly, the GEM cnsiders the cntext fr the idim and hw it helps the cmprehender understand that a literal interpretatin is inapprpriate, and als hw it helps by semantically prviding a pssible figurative meaning (Oakhill, Cain, & Nesi, 2016). The main questins the GEM cncerns are (a) Hw and when the child exceeds this lcal interpretatin and prcesses figuratively, and (b) What cgnitive and linguistic abilities the child needs in term f figurative language, in rder t prcess nn-literally. These abilities are defined by Levrat and Cacciari (2002) thus: 10

19 1. The ability t cmprehend the dminant as well as the peripheral meanings f a wrd, and its psitin in a given semantic dmain; 2. The ability t g beynd a literal strategy f language interpretatin. This is a prerequisite necessary t cpe with mst f the linguistic repertire nt nly with figurative language; 3. The ability t use cntextual infrmatin t cnstruct a cherent semantic representatin f a text by integrating different surces f infrmatin; 4. The awareness that what is said and what is meant des nt always cincide. (p. 129) The GEM ges n t identify five develpmental phases. The first f which is a primitive prcessing where the child interprets language literally in all cases. The secnd phase, apprximately at the age f eight t nine, the child realises that there is a discrepancy between what is said and what is actually cmmunicated n the basis f cntext. In phase three children recgnize that the same cmmunicative intentin can be reached thrugh different sentence frm (literally, metaphrically, idimatically, and s frth). Ten t twelve year lds take int cnsideratin the intentins f the speaker and their wn knwledge. By phase 4 they can ften use and cmprehend a cnventinal repertire f expressins nrmative fr fifteen-year-lds. The final phase is the adult-like figurative cmpetence, generally cnnected with the ability t use figurative expressins in a creative way, based n metalinguistic and metasemantic awareness (Levrat & Cacciari, 2002). The hypthesis f such a stretched develpment is supprted by Nippld and Taylr (1995) wh fund that adlescents struggle with idimatic expressins withut sufficient cntext. In a study n Bulgarian children with Bulgarian as L1, Vulchanva et al. (2011) fund that cmprehensin f figurative language crrelates with age and years f schling. They als cnfirmed that it was linked t metalinguistic awareness and the ability t infer frm cntext. In line with the GEM, they fund that age 10 was a turning pint in idimatic knwledge fr Bulgarian L1 speakers. Of interest, Vulchanva et al. equipped a typlgy fr idims nt much used in previus studies f similar scpe, in rder t address the absence f a standard idim typlgy fr Bulgarian. The typlgy is rted in human experience, assuming that idims either have a bilgical r cultural surce, and a nvel categry which they dubbed instructive idims. Bilgical idims, scring highest fr ten-year-lds and adults, build n analgies invlving parts f the human bdy (fr example get n NP s nerves, save NP s skin ). The success f this categry, they say, is that they lexicalize the basis f human experience, and that they are shared acrss cultures and acrss the Ind-Eurpean languages. They further fund that, measuring these idim categries against each ther, while the 11

20 bilgical idims scred highest in cmpetence, the instructive idims placed secnd and the cultural last. The instructive idims, hwever, are typically very cmpsitinal, lking very much like sayings and prverbs. What Vulchanva et al. cncluded n this was that their interpretatin allws fr the ability t infer frm minimal cntext. This falls in line with Vega- Mren s view n prcessing unfamiliar and nvel idims. Sme idims prved mre difficult than thers in their results, bslete lexicalisatin, and especially idims based n bslete grammar r syntax prved difficult t prcess fr the Bulgarian children. As fr that particular study s results, there was n clear indicatins fr parallel prcessing, because incrrect answers primarily ccurred with literal interpretatins, and they had n measure fr prcessing speed. Oakhill et al. (2016) studied skilled and less skilled cmprehenders frm tw age grups, 8- and 10-year-lds, fr understanding f idimatic expressins in cntext. They evaluated nline prcessing and interpretatin by measuring reading times and cmpetence with shrt texts cntaining English idims and nvel, translated Italian idims. As expected the lder children and better cmprehenders were mre likely t interpret idimatic phrases crrectly than the yunger children. The reading time results did turn up with sme interesting results, hwever. The less skilled cmprehenders spent less time when reading the nvel idims than did the mre skilled cmprehenders. This indicated that the gd cmprehenders were mre likely t attempt an idimatic prcessing, which takes lnger time than a literal ne. The study fund that lnger reading times crrelated with a better cmprehensin f idimatic expressins, particularly unfamiliar nes. The cntext in the stries allwed fr equipping inferential skills and metaphrical mappings, clearly separating the gd frm the bad cmprehenders, as well as shwing the difference between the age grups in line with the GEM. Levrat and Cacciari (2002) cnducted tw parallel experiments n metalinguistic awareness and its influence n the ability t cmprehend and prduce figurative expressins, funding n the GEM. The first study targeted develpment f metalinguistic awareness thrugh an elicitatin task, while the ther tested the newly created expressins fr apprpriateness, cmprehensibility and nvelty with adult judges. Interestingly, in the first experiment they fund that the ability t create new expressins with figurative language increased frm childhd t adlescence but nt frm adlescence t adulthd. In the secnd study they fund that adults creatin f nvel expressins were judged as mre apprpriate 12

21 than thse f the adlescents. This supprts the predictin in the GEM that metalinguistic abilities cntinue t develp thrugh and after adlescence. Anther interesting study investigated mental imagery in relatin t the cmprehensin f transparent versus paque idims. Many previus studies mentined here have an apprach and scpe twards the cmpsitinality f idims, but this particular study by Nippld and Duthie (2003) suggested that the cmprehender s mental imagery als underges a develpmental prcess. Building n an apprach that idims are partly learned by analysing the wrds cmpsing them, the Metasemantical hypthesis, favrs transparent expressins. They hypthesised that the mental imagery is key fr understanding idimatic expressins and mentin tw idims t explain this; upn hearing paddle my wn cane in a cntext, the cmprehender might picture smene maneuvering a large cane withut assistance, giving him clues which tgether with the cntextual clues lead t an understanding. Hwever, with paque idims like paint the twn red the mental imagery f smene actually painting buildings might nt serve the understanding f the intended meaning very well. Cntrary, the mental image might mislead him if he has never befre encuntered the particular idim. The results f this study suggested that as adults had mre mental imagery related t idims than yunger subjects, it culd mean that the mental imagery relates t a prelearned knwledge helping thse wh knw the expressins already. The children had less idimatic imagery because it tends t mislead and des nt help them acquire new transparent idims. On this basis, Nippld and Duthie (2003) cncluded that the mental imagery f idims might serve as a barmeter fr their depth f understanding idimatic expressins Idims in Secnd Language Acquisitin Secnd language learners meet figurative expressins in all frms f discurse, and it is a part f the Nrwegian English L2 curriculum. Yet L2 learners are at a disadvantage in understanding figurative expressins in their L2, much because they are at a lwer level f cmpetence in the L2 cmpared t their L1.Research n L1 idim cmprehensin is extensive and mainly fcuses n English native speakers. Mst f the research n L2 idim cmprehensin is influenced by the ntin f crss-linguistic influence, als knwn as transfer, which Odlin (2003) characterises as the influence resulting frm the similarities and differences between the target language and any ther language that has been previusly (and perhaps imperfectly) acquired (p. 436). Transfer is influenced by scial setting, as all 13

22 language acquisitin take place in a scial matrix. Thus there are certain imprtant factrs t cnsider cncerning L2 speakers cmprehensin f figurative expressins in the target language. Studies have shwn that the hetergeneity f speakers, fr instance culture, scial class and gegraphy matters in what influence languages have n each ther (Odlin, 2003). The linguistic relativity between languages leads t a cnceptual transfer, the travel f previusly unfamiliar cncepts frm ne language t anther. An interesting example n this travel is the study by Pavlenk (1999) which subjected different grups f Russians with English as their L2, as well as Russian mnlinguals, t a film with the aim t make them talk abut the cncept f privacy. Russian language has n equivalent cncept r wrd fr privacy. The study shwed that the Russian immigrants living in America described an equivalent with mre accuracy than did thse wh lived in Russia. The resulting Russian phrase fr privacy turned ut t be a hybrid between English and Russian. Studies n the rle f L1 n L2 idim prcessing shws that the L1 influences prcessing, even thugh L2 learners are less likely t transfer L1 knwledge when encuntering figurative language. This indicates that althugh L1 is influential, there must be ther factrs in play fr L2 idim prcessing (Cper, 1999). The critical difference in idim prcessing between the L1 and L2 speaker f a target language, is the time they spend prcessing it. As native speakers usually react in a split secnd t what is usually a familiar idim, the situatin is different fr the L2 speaker wh will usually take time t screen different prcessing strategies. The latter has t take int cnsideratin many pssible answers and slutins thrugh trial and errr. In a qualitative study, Cper fund that L2 speakers used a heuristic apprach, attempting varius strategies t find an idimatic meaning fr what Cper call a linguistic prblem. He argues that the mdels fr idim cmprehensin, mst f which are already mentined in this framewrk; The idim list hypthesis, the lexical representatin hypthesis, the direct access mdel and the cmpsitin mdel are all t limited in scpe t accunt fr the variety f strategies equipped by L2 speakers in Cper s study. He sums the strategies used int tw grups: One which was guessing frm cntext, using the literal meaning t understand the figurative meaning and discussing analysing the idim in rder t gain necessary knwledge f it. The ther grup, emplyed significantly less than the first, was requesting infrmatin, using backgrund knwledge, referring t an L1 idim and ther strategies (Cper, 1999, p. 255). The reasns fr the small amunt f research int L2 idim acquisitin culd stem frm the belief that idims are mre like an auxiliary part f language, that it cmes with expsure, 14

23 r frm the ntin that English is especially idimatic. The latter is wrng, studies cmparing English, fr instance with Spanish, shw that they are equally idimatic (Styanva, 2013). Few studies cnsider tw f the cnsequences f transfer, avidance and hypercrrectin, and even fewer cnsiders the imprtance f culture in studies n L2 idim prcessing (Styanva, 2013). It seems a cnnectin between secnd language acquisitin and figurative cmpetence is a difficult matter, and it lacks research. There are factrs, besides the transfer phenmenn, that can prvide meaningful infrmatin in this cnnectin, and they are 1) Degree f expsure t the idimatic expressins, and 2) Metalinguistic awareness Degree f Expsure Studies investigating this are usually experimental, testing different strategies n students in rder t see which is mre effective in teaching L2 idims. As with all secnd language acquisitin, idim acquisitin relies n familiarity either with the expressin itself r with the system fr the expressins in general (much discussed in the sectin n representatin and prcessing f idims). Accrding t Ellis (2002), the frequency f a phrase r wrd in the input determines the grwth f this lexical item r phrase. Lexical items and phrases that ccur mre frequently have a higher prbability f being recgnized and prcessed in general. This is in line with the hyptheses hlding that familiar idims are easier t prcess. Nippld and Rudzinski als emphasize practice fr ptimal prcessing (1993).An interesting example is the study by Eve Zyzik (2010). She hlds that althugh idims are best acquired thrugh interactin with native speakers, explicit teaching with idims in L2 acts as a gd stepping stne fr building awareness f their existence and hw they are used in discurse. Zyzik studied 65 English native speakers fr their Spanish L2 idims. The experiment was divided int grups, ne grup was taught idims gruped thematically. The ther grup was taught idims withut thematic fcus, but was tld t fcus n the main verb. The results fr this experiment was that the students, regardless f strategy, imprved in terms f prductin and recgnitin f idims. Of nte, the students imprved much mre in terms f recgnitin than f prductin. The imprtance f expsure t idims in secnd language acquisitin is well dcumented. Since idims vary in transparency, explicit teaching in terms f etymlgical elabratin is beneficial when it cmes t cmprehending and acquiring unfamiliar nn-transparent idims. Nn-transparent idims are prblematic because they are unique t their native language and 15

24 have underlying rigins that may explain their meaning (Styanva, 2013). This is als a tpic in the cming sectin n metalinguistic awareness Metalinguistic awareness In this field, Metalinguistic awareness is frequently distinguished frm Epilinguistic capacity (als knwn as tacit knwledge ). Epilinguistic capacity is the prcess f mnitring language t crrect lexical and phnlgical errrs, while metalinguistic awareness refers, as Gass and Selinker (2001) put it, t ne s ability t cnsider language nt just as a means f expressing ideas r cmmunicating with thers, but als as an bject f inquiry (p. 302). Metalinguistic awareness can be analysed by three layers: Wrd awareness, Syntactic awareness and Pragmatic awareness (Benelli, Belacchi, Gini, & Lucangeli, 2006). Benelli et al. hlds that metalinguistic ability crrelates with cgnitive develpment, metacgnitin, literacy and ral language skills, and that it has been fund t develp with schling (2006). This indicates that metalinguistic awareness is clsely cnnected with the expsure factr in terms f idim acquisitin. Nippld (2006) states three metalinguistic strategies fr language acquisitin: Cntextual abstractin, Metalexical analysis and Wrld experience. Using the cntextual abstractin strategy, the learner infers meaning frm the linguistic cntext f the expressin. In the metalexical strategy the learner infers meaning frm the lexical items in the expressin, while the wrld experience refers t the learner s experience and expsure f figurative expressins (Nippld, 2006). The latter can be interpreted as the learner s familiarity f the given expressin, but is als a package f metalinguistic abilities like fr instance the ability t understand etymlgical backgrunds and cultural aspects f the target language. As has been shwn earlier in this framewrk, idimatic cmprehensin crrelates with metalinguistic awareness, which in turn develps with expsure and years f schling (Nippld, 2006; Vulchanva et al., 2011). 16

25 3. Methds and Prcedures The aim f this study was t investigate upper secndary students cmprehensin f idims in their secnd language, and try t analyse their results using theries fr prcessing idimatic expressins. Twenty-nine upper secndary students were given tests t measure their cmprehensin f idims and their verall English language prficiency. The idim test was divided int tw parts: One part cntained 30 idim questins each presented with a shrt written cntext and fur alternatives, the ther part cntained 30 idim questins als with a shrt written cntext and fur alternatives, but n a MuseTracker platfrm. In this latter part we had fur drawn images as alternatives fr each idim questin, and the platfrm allwed us t track their reactin times and muse-mvement trajectries. A grup f advanced students f English were als measured fr idim cmprehensin. 3.1 Participants Three sets f participants were part f this study. As mentined earlier, twenty-nine Nrwegian upper secndary students. 18 f them were females, 11 males, all aged between 16 t 17 years. This grup cnsisted f a single Vg1 class with English as an bligatry subject. All the students f this class were encuraged t participate, hwever, three students had t be excluded frm the study due t nt cmpleting all tests. Analysing the backgrund infrmatin sheets (Appendix 2), we cncluded that the remaining 26 participants were eligible fr the study. The secnd set f participants cnsisted f twenty-six students frm NTNU, all advanced learners f English picked frm English study prgrams between secnd and fifth year in prgressin. Of these were 17 females and 9 males, aged 20 t 28 years with a mean age f 24. In line with the guidelines f the Nrwegian Scial Science Data Service (NSD), the participants had t sign a frm f cnsent befre participating in the study. The students were infrmed abut the cntent f this paper and the tests befre signing. Eight participants reprted t have ne r mre deficiencies r diagnses that may affect ne s learning ability like Dyslexia, Attentin Deficit Disrder, auditry r visual impairments. Nne f these participants results deviated frm the nrmal distributin f the tests cnducted. We decided nt t remve any f these participants based n the results. NSD apprved the research prject, and in line with their guidelines we never btained any names. The students frm 17

26 bth grups were given participant numbers, which identified the data cllected frm the backgrund survey with the idims and English language tests The native speakers The third grup f participants was the English native speakers. This grup was used as a way t certify that the idims we picked and the written cntext fr them were based n knwn, understandable idims in the English language, as judged by English L1 speakers. Of these 109 we had t exclude 10 participants n grunds f them either nt being native speakers f English r being extreme utliers. Thus they cnsisted f 99 participants. We did nt cllect any backgrund infrmatin frm this grup. 3.2 Materials and Prcedures The English Prficiency tests The upper secndary grup was suggested by ur supervisr wh has had cntact with that particular schl fr ther prjects. The teachers there were familiar with the type f research cming frm ur department, and a teacher was willing t give us time with a class. The NTNU student grup was gathered frm advanced English curses at NTNU Dragvll campus. The first thing we wuld test after filling ut backgrund schemas and papers f cnsent, were the English vcabulary and grammar tests. The results frm these helped in establishing a baseline fr English prficiency in rder t establish a measure fr cmparisn. The vcabulary test cnsisted f 100 tasks presented as letter sequences (vcabulary.ugent.be, 2015). Sme f them were existing English wrds and sme were made up nn-wrds. The participants wuld respnd with a yes r a n t each sequence, judging whether they thught it was a real wrd r a nn-wrd. The vcabulary size was estimated after the test and wuld be displayed n the participant s screen. The grammar test we used was btained frm Cambridge s English prficiency tests (examenglish.cm). This test presented 50 tasks with 4 alternatives in each. The questins prvided a sentence with a missing wrd. The crrect respnse wuld be amng the 4 alternatives, and if the participants selected a wrng respnse, they wuld have t try again until they reached the crrect ne. Bth these tests were advanced and suited fr native speakers, s the difficulty was high. We wanted this s that we culd easier cmpare ur tw grups with each ther. Upn cmpleting a test, we wuld persnally check the scre and write it dwn fr them befre mving further with testing The Idim tests These tests prved t take very lng time, due t a number f reasns. We needed a sufficient 18

27 amunt f idims with respnse alternatives, a cntext fr all the idims, and suitable pictures fr half f the idims. The test itself cmprised f a ttal 60 idims. We selected idims frm varius surces, mst f them were simply fund by ggling fr idims. Hwever, we had t be critical and selective when deciding which idims t use fr the tests. We decided against idims that seemed either t easy r t difficult, as judged by urselves. Our supervisrs helped in this prcess as well, by selecting idims frm ur list. We made a pint f including different kinds f idims, frm transparent t paque, and fllwed typlgies like the cmpsitinal variances discussed in Titne and Cnnine (1994). Thus we included Nrmally decmpsable idims, like caught red-handed, in which part f the idim is used literally. Abnrmally decmpsable idims, like give smene a hand, in which the referents f an idim s parts can be identified metaphrically. Semantically decmpsable idims, like Devil s advcate, in which the meaning cannt be derived frm the literal prcessing f the string. Als the typlgy described in Vulchanva et al. (2011) was equipped. We included instructive idims, like dn t put all yur eggs in ne basket, bilgically mtivated idims, like lse yur head and culturally mtivated idims, like kick the bucket. Hwever, it was never the intentin t cntrl the study fr typlgy specifically. It is clear by the typlgy frm Vulchanva et al. that all three f these categries may vary acrss cmpsitinal dimensins as well. The bilgical and cultural smetimes melt tgether. The idims were selected in general because they are used in English language by native speakers. The reasn fr picking idims frm different categries was grunded in the cmplex nature f idims, and the studies directed at mapping the different effects these have n prcessing them. The 60 idims were split in half, ne written frm and ne picture frm. The reasn fr this is is twfld. Fr ne, the cntext in which the idimatic expressin is given is significant, accrding t the Glbal Elabratin Mdel (GEM), attentin t the cntext allws the cmprehender t realise that a literal interpretatin f an idimatic expressin is inapprpriate. Furthermre, it gives the cmprehender necessary semantic infrmatin in rder t reach an apprpriate figurative meaning frm the expressin (Levrat & Cacciari, 1995). This is als stressed in Vega-Mren s discussin n acquiring unfamiliar idims, she suggested that the cntext tgether with the encded material that frms the string is a starting pint fr pragmatic inference (2001). There is evidence that the mre skilled cmprehenders gain mre advantage frm the cntext than d the less skilled cmprehenders, and evidence that cntext helps lder children mre than yunger children (Levrat & Cacciari, 1995). We 19

28 did nt cnsider these factrs as prblems, because ur subjects were adlescents and adults fr which there is n such cntext research. The English prficiency tests wuld reveal any big deficiencies between skilled and less skilled cmprehenders. We assumed that the cntext wuld prvide a degree f predictability, increasing the prbability f cmprehending a figurative expressin crrectly. The dimensin f familiarity was interesting fr the tpic f this thesis, but due t the limited scpe we culd nt test fr this cnditin specifically. Instead we chse the idims with a belief that they wuld represent bth familiar and unfamiliar expressins fr the upper secndary students. Fr the advanced English students, it wuld be likely that nt all f the 60 idims were familiar. Because the participants were Nrwegian students with English as L2, it was likely that they had nt achieved a native-like prficiency in English. This is especially true fr metalinguistic abilities, which accrding t the GEM develp thrugh and beynd adlescence. Familiarity has, as mentined in the theretical framewrk, been shwn t influence cmprehensin f idims (Titne & Cnnine, 1994) and influence prcessing (Vega-Mren, 2001). Under this dimensin we als cnsidered the cncept f crsslinguistic influence. Many f the idims are used even in Nrwegian L1 discurse, like fr instance Blckbuster and Brken heart, either in their riginal English frm r in a Nrwegian versin. The first half f the idims test, cntaining 30 idims, was given t bth the student grups as well as the native speakers. They were set with a brief stry cntext intrducing the idim inside it. An example f cntext and alternatives fr ne f these tasks was: They all thught the match wuld be a piece f cake but they were wrng. The ther team was faster. What is the meaning f the expressin piece f cake? 1. T be satisfactry 2. The reward yu get fr winning 3. Smething which is easy 4. The best part f the cake t eat The alternatives are designed s that tw alternatives are distractrs, and cmpletely wrng, like 1 and 2 abve. One is crrect, and ne is clsely related t the idim in questin literally, like 4 abve. The tw wrng alternatives were in mst instances far ff frm the crrect answer, but were in ther tasks smewhat related t the cntext sentence. This was the system 20

29 thrughut ur idim tasks. The intentin with the literal alternative is related t the theries f prcessing figurative language. Althugh the participants in this study were frm adlescence and abve, they culd be tempted t rely n a literal interpretatin presented t them if they met with an unfamiliar idim. If they chse the literal answer, it culd indicate that they chse a path f literal prcessing, either aviding the pssibility f figurative interpretatins r fund them inapprpriate. The literal alternative als has a rle as an indicatr twards the figurative interpretatin. Having cmpleted mre f the idim tasks, the participants are likely t realise that the literal alternative is inapprpriate, and lk fr figurative interpretatins beynd literal meaning. We fund that this system wuld generate interesting data, and the gal was t make the tests neither t difficult nr t easy fr the tw grups. The tests were cnstructed by us, and submitted by the participants n the Ggle Frms platfrm. The questins came in randmized rder fr every participant. The ther half f the idims had a different design. Instead f the written alternatives mentined abve, we included fur images fr each task. Fr this we made drawings which we in turn scanned. An example f such a task lks like Figure 1. Figure 1. Idim task with image alternatives. Tim was feeling tired. He decided t hit the hay. Chse the picture that shws the meaning f "t hit the hay" The stimuli in these tasks is different frm the ther 30 tasks. Nt nly des the cntext give a clue, but the images allw the participants t make an inference based n the cntent f the 21

30 image alternatives. Nippld and Rudzinski (1993) accmmdates the idea that idims can be prcessed by a variety f strategies, and that the availability f cntext clues is amng them. Nippld and Duthie (2003) studied whether mental imagery play a rle in acquisitin f idims, and fund that it des fr bth transparent and paque idims. The rle f mental imagery was in ur mind as we made the pictures, thus the fcus was t make them as generic and familiar as pssible and nt leave any rm fr ambiguity. The ther reasn fr the use f images was t fit the platfrm we used t btain data frm the upper secndary grup. This platfrm is the MuseTracker sftware. It is a free prgram fr studying real-time mental prcessing using a muse-tracking cmputer methd (J. Freeman & Ambady, 2010). The design fr this test, the 30 idims chsen fr images, was chsen fr the upper secndary students. We wrked n making cdes in the MuseTracker prgram s that ur test wuld run ptimally. The way it wrked was that the participants wuld use ne f ur three cmputers, enter their participant cde and begin the test. We gave them instructins n hw t navigate the test and what they culd expect frm the tasks. Each task gave the participant first the cntext sentence and the questin f what the idim meant. They then were presented fur images, such as thse shwed abve. The images were put each in ne crner, and the muse wuld always begin in the exact middle f the screen. As sn as they pressed start, the prgram began recrding every mvement f the muse, the streaming x- and y- crdinates and the time spent alng them. The participants received the tasks in randmized rder, and the respnse alternatives were als randmized fr each task. This had ne exceptin, the literal and the crrect alternatives wuld always be lcated next t each ther hrizntally. The reasn fr this is that we needed t be able t analyse the respnse mvement in between the literal and the crrect mre than the thers. Accrding t Freeman, Dale and Farmer, sme cntexts f language prcessing leads t a nt s smth prcessing flw, which in turn leads t hand mvements that exhibit sharp shifts in directin (2011). If such a disruptin between the literal and crrect alternatives were t happen, that culd tell us smething abut the participant s prcessing, particularly whether the participant shifted between a literal and figurative interpretatin, in accrdance with (Oakhill, Cain, & Nesi, 2016; Vega-Mren, 2001). Using the MuseTracker sftware has a number f benefits. Althugh it is nt as ppular as the eye-tracker, it can prduce gd data in a different manner. Freeman suggests the MuseTracker s perks and usefulness in his article Hand in mtin reveals mind in mtin (2011) The dynamics f actin d nt reside nly in the aftermath f cgnitin, bdily 22

31 mvements such as traces f hand mvements can prvide insight t internal cgnitive prcesses. The distinctin between eye and hand mvements is that the latter ffers a cntinuus stream f mtr utput whereas the frmer typically is cmprised f discrete saccades (J. B. Freeman et al., 2011). Freeman further suggests several methds f data cllectin frm this type f test, including maximum deviatin tward unselected respnses, switches in directin and mvement cmplexity. T sum up this sectin n preparatin fr data cllectin: Tw main grups f respectively upper secndary students and university students f English were tested fr 60 idims and English prficiency. The idims were presented in tw cnditins, with written cntext and picture cntext. The upper secndary students perfrmed the picture cntext idims n a MuseTracker platfrm, while the university students perfrmed them n a Ggle Frms platfrm. A third grup f native English speakers received the 60 idims n a Qualtrics questinnaire platfrm. Their results was nt part f the cmparative analysis, but was used as a baseline fr the quality f ur idim tests. 3.3 Analysis All the MuseTracker data was retrieved frm the MuseTracker Analyser sftware and then run in SPSS. The participant data and the idim task data was then run in binmial tables (/binmial (0.50)=errr) t detect results that fall utside nrmality. The accuracy data was cllected frm this sftware as well as frm the written tests and image tests cnducted n nline frms. We prceeded t add these results up in Excel which gave us a summary f hw many respnses bth grups made fr the three cnditins: Crrect, Literal and Distractr. This data was then run in R, where we first tested them fr nrmality bth within-grup and between grups. Hwever, because the university students perfrmed at ceiling, we needed a mre rbust methd t cmpare these data. T measure the tw grups against eachther fr three different cnditins we chse t run a repeated measures analysis f variance (ANOVA) in R. We added all the data frm bth grups in a text dcument, listed as bservatins. Every participant frm bth grups wuld then have tw bservatins each, ne frm image tests and ne frm written tests, these made the experimental factrs f the ANOVA, which cmputed a between-subjects variable. The dependent variable in the ANOVA mdels was either Crrect, Distractr r Literal. The tw stimulus types were Image and Written, referring t the type f idim tasks. It was imprtant t distinguish these tw types in the ANOVA, s we culd see the difference in perfrmance between the tw stimuli. 23

32 3.3.1 The English Prficiency tests The methds f analysis differed acrss the type f tests cnducted in this study. Fr the English prficiency tests, the results had t be numerically cded fr cmparisn. While the vcabulary test gave a percentage base fr hw many wrds it estimated the participant t knw, the grammatical test wuld give a pint scre between zer and fifty. We adjusted the scales t g between zer and ne hundred. We intrduced these data int excel sheets and prceeded t run them in R fr analysis The idim tests Having first sent all the idims in a Qualtrics frm t English native speakers, we received data which helped us decide if any items were t be excluded. The native speakers generally answered crrectly, but 7 idims did nt wrk ut well. These 7 idims, marked with asterisks in the [APPENDIX] had nly between 40% and 69% success rate. We assumed this meant that the native speakers either did nt knw them well, r that the frmulatin f the tasks were bad in sme manner. These idims culd be excluded, but in this analysis I decided nt t exclude them. The reasn fr this is that the three grups had different results with them. The university students in general did nt have many errrs at all, and this was als true fr the idim tasks that the natives failed with. The upper secndary students did in fact have mre errrs in a few f the idim tasks that the natives als failed, but it was difficult t see a crrelatin. Nevertheless, the scpe f this study is t investigate the prcessing f idims, and the respnses made by the upper secndary students wuld be interesting in any case. The MuseTracker data prvided the mst difficult target fr analysis. Fr every task, each participant respnded with a muse mvement tward ne f fur alternatives. The mvement was recrded times per secnd, measuring the speed, time spent and crdinates fr the muse pinter. This amunts t very much data with the ttal f 780 respnses we received. S fr the results chapter, we have made a strategic decisin by remapping the trajectries. Thus all the visual results frm the MuseTracker are distinguished and treated as shwn in Figures 2 and 3. Figure 2. Example f raw MuseTracker trajectry data. 24

33 Figure 3. Example f remapped MuseTracker trajectry data. Nte: T the left is crrect respnses, t the right is incrrect respnses. The remapping nly applies t crrect respnses. There was unfrtunately n way f remapping incrrect answers t highlight the distinctin f literal versus distractr respnses. Als nte that these figures shw a ttal f 750 respnses instead f 780. This is misleading, as ne participant s respnses culd nt be prjected due t all that participant s respnses being crrect. 3.4 Ptential Surces f Errr and Limitatins The selectin f idims fr the tests is the greatest limitatin fr this study. As mentined earlier, the scpe f this study did nt allw us t accunt much fr the familiarity f the idims. It is difficult, hwever, t make an accurate calculatin f an idim s frequency, and this is prbably especially true fr the L2 speakers. Secnd, as Jackendff (2002) and Vega- Mren (2001) stressed, the idim flexibility must be cnsidered. This limits the scpe f frequency even mre. Which variatins d the natives accept, and which d the English L2 speakers knw and cmprehend. The nature f the flexible idims and their cmplex variatin is discussed in the theretical framewrk, but a few examples f decisins we made will be presented here. Fr example, in the idim speak f the devil, the verb speak culd be replaced with the verb talk withut changing the expressin s figurative meaning. Anther example is the idim give smene a hand, which culd be changed structurally by adding 25

34 a prepsitinal phrase and mving the nun phrase t the end, creating the expressin give a hand t smebdy. Secnd, the idim tasks judged prly by the native speakers culd in fact mean that thse particular tasks were bad. It is uncertain t us why the native speakers failed them s distinctly, while the Nrwegian university students did nt. This will be thrughly discussed in a sectin in the discussin chapter. Fr the sake f analysis I chse t keep them and this must stand as a factr fr ptential weakness f the study. Last, the different data cllectin frm the tw main grups was different. Fr a study cmparing tw grups, the mst valuable data cmes frm the mst identical data cllectin pssible. The upper secndary students mitted the idim image results thrugh MuseTracker sftware, while the university students did the ggle frms survey fr all the idims. Hwever, due t the scpe f this study and the limited amunt f resurces, we agreed with ur supervisrs that cnducting the test in this manner was the best slutin. Fr this paper, lking at the prcessing f idimatic expressins with the upper secndary students as main fcus, this data shws value. Despite this and ther weaknesses, the carefully planned data cllectin was assumed t prvide sufficient data. This will be further included in the discussin f the results. 26

35 4. Results This sectin will describe the results f the prficiency tests as well as the idim tests. The mst valuable aspects f the data sets will be presented and analysed. Reasns fr why thse particular aspects f the data were chsen fr analysis will als be presented. The data was analysed in MuseTracker analyser, R and SPSS. A repeated measures analysis f variance (ANOVA) was used t cmpare accuracy in the tw grups in the idim tasks. The MuseTracker data was retrieved frm the MuseTracker Analyser and treated in SPSS afterwards. In the analysis prcess, three different perfrmance variables were cnsidered. In sectin 4.2 accuracy is presented. The tw grups were bth measured fr hw many crrect, incrrect and literal respnses they made. In sectin 4.3 and 4.4 the data frm the MuseTracker sftware is presented. First the reactin times fr the image idims frm the upper secndary grup. The last data frm MuseTracker, in 4.4, will cncern the certainty f the muse mvements fr the upper secndary grup fr the same test, the image idims. 4.1 The English Prficiency Tests The prficiency tests cnducted n bth the main grups serves as a baseline fr cmparisn. Table 1. Grammar and Vcabulary tests by Mean and Standard Deviatin Grup N Mean Std. Deviatin University Vcabulary Grammar Valid N (listwise) 26 Upper Secndary Vcabulary Grammar Valid N (listwise) 26 Nte: University = Nrwegian university students. Vcabulary = The estimated size f English wrds in the vcabulary by percentage, Grammar = Pints ut f 100 pssible calculated frm crrect answers. Since a cmparisn between the tw grups is interesting, a test was run between the tw grups fr significance. A wilcx test between the tw grups fr the vcabulary data shwed an effect value f 68 and p <.001. The grammar test als shwed significance with an effect value f 26,5 and p <.001. The vcabulary test shwed a high scre fr the university students. It estimated their vcabulary size t be 63% f the English wrds. The upper secndary students, hwever, 27

36 achieved a much lwer scre f 39%. In the grammar test we see n Table 1 abve, the university students achieved 82,5% f the pssible pints, while the upper secndary achieved 56,4%. The difference between the tw grups in vcabulary = ~24 percentage pints, while fr grammar = ~26 percentage pints. Nte that the spread between subjects in the tw grups differ. The standard deviatin fr the upper secndary grup is ver 15 pints n the vcabulary test, cntrary t the 10 pints fr the university students. The high spread fr the upper secndary grup reflected the clear difference in perfrmance by that grup. 4.2 Idim tests: Accuracy Table 2. Idim respnse accuracy Grup N Mean Std. Deviatin University students Image Crrect Literal Distractr Valid N (listwise) Written Crrect Literal Distractr Valid N (listwise) Upper secndary students Image Crrect Literal Distractr Valid N (listwise) Written Crrect Literal Distractr Valid N (listwise) 26 28

37 Nte: The Mean values are measured in percentage. These results shw the tendency frm the prficiency test in sectin 4.1 in that the university students utperfrm the upper secndary students clearly. Remarkably, the standard deviatin fr crrect respnses in the upper secndary grup is ~20.5. As seen in Figure 4 belw, the spread between participants was much bigger in the upper secndary grup (dented as HS in the figure) than in the university students. Figures 4 and 5. Bxplt f Crrect respnses t the left and Distractr respnses t the right, fr bth grups by bth stimuli. Nte: HS = upper secndary grup. Uni= university student grup. The scale frm 0 t 100 dente percentage f crrectness.. Figure 6. Bxplt f Literal respnses fr bth grups by bth stimuli. 29

38 These figures (4, 5 and 6) shw that the university grup perfrms at ceiling. Their data are nt distributed nrmally as in the upper secndary grup indicated by this ceiling effect. Bth cnditins, crrect and incrrect, shw that the written stimuli was mre difficult fr bth grups, particularly the upper secndary grup. The results frm the ANOVA illustrated in these bxplts shw that there is a difference within-subjects and between grups fr the varius cnditins. This is shwn in Table 3 belw. Table 3. ANOVA. Crrectness Df Sum Sq Sq Mean F value Pr(>F) (*) Grup E-10 Residuals Stimulus E-05 Stimulus:Grup Residuals Literals Df Sum Sq Sq Mean F value Pr(>F) (*) Grup E-05 Residuals Stimulus Stimulus:Grup Residuals Distractrs Df Sum Sq Sq Mean F value Pr(>F) (*) Grup E-13 Residuals Stimulus E-07 Stimulus:Grup E-07 Residuals

39 4.3 Idim test: Reactin Times Table 4. Reactin Times by Mean and Standard Deviatin Errr N Minimu m Maximu 0 RT Valid N (listwise ) 1 RT m 0 0 Mean Std. Deviatin Valid N 181 (listwise ) Nte: Errr 0 = Crrect answers, Errr 1 = Incrrect and literal answers. RT is measured in millisecnds. Valid = Respnses SPSS wuld cnsider, in this case all f them. As illustrated in Table 3, the reactin times are generally higher fr the incrrect respnses than fr the crrect respnses. In fact, the difference by mean RT is twice the size fr incrrect respnses. The mean fr crrect respnses is ~2 secnds, while fr incrrect respnses it is ~4 secnds. The results shw a tendency tward shrter RT fr mre crrect idims, and lnger RT fr the mre incrrect idims. Idims like Bent ut f shape (61,5% crrect), Cst an arm and a leg (73% crrect), Put all eggs in ne basket (53,8% crrect) cmprise the three highest mean RT s. Idims like Hit the hay (92,3% crrect), Piece f cake (100% crrect) and Bite ver mre than yu can chew (88,5% crrect) cmprise the three lwest mean RT s, and are amng the mst crrect idims fund in the image idims test. 31

40 Figure 6. Reactin Times per Item Nte: The cnditins listed are listed as abbreviatins. See appendix fr translatin f these. Generally they are easy t decde, fr instance bat = t be in the same bat, rain = it s raining cats and dgs, bird = t kill tw birds with ne stne. The mean RT is measured in millisecnds. 4.4 Idim test: Respnse Certainty The final factr fr analysis was the certainty f the decisins made by the upper secndary grup in the MuseTracker sftware. Accrding t Freeman and Ambady (2010), ne f the best methds fr measuring certainty in MuseTracker is t lk at the participant s attractin twards an unselected target in a given task. Tw measures fr this, Maximum Deviatin (MD) and Area Under the Curve (AUC) are used. The sftware cmputes an idealized (straight line) respnse trajectry between the start and endpints fr the respnses. The MD f a trajectry is the largest deviatin between this idealized (straight line) trajectry and the actual trajectry. A high MD then means a high attractin tward an unselected alternative. The AUC f a trajectry is calculated frm the area between the idealized and the actual trajectry, while the area n the ppsite side f the idealized trajectry is calculated as negative area. Freeman and Ambady (2010) bserved that the AUC is a better index f the verall attractin tward the unselected alternative (incrprating all time-steps), whereas MD 32

41 is a better index fr maximum attractin, but this attractin may be limited t fewer time steps (p.230) Figure 7. Maximum Deviatin in crrect respnses. Figure 8. Maximum Deviatin in Incrrect respnses. Nte: Mvement in either directin away frm the 0 value equals deviatin. This means that bth psitive and negative values indicate a deviatin. The values are gemetric crdinates within the MuseTracker trajectry space. 33

42 As these tables illustrates, the respnses resulting in crrect answers have far smaller MD values. This indicates that the trajectries, the muse mvements, are a great deal mre certain twards the target when the participant is answering crrectly. A ntable idim in Figure 7 is tie the knt (cded as knt in Figure 7), fr which the crrect respnses seem t be mre uncertain than the incrrect nes in Figure 8. As fr the AUC, illustrated in Figures 8 and 9, a similar trend was fund. Figure 9. Area Under the Curve in crrect respnses. Figure 10. Area Under the Curve in incrrect respnses. 34

43 Nte: Idims with 0 errrs, like Piece f Cake (cded as cake in these figures) are excluded. 35

44

45 5. Discussin The aim f this study was t investigate idim prcessing in Nrwegian L2 speakers f English. The grup f interest was a class f 16 t 17 year-ld adlescents wh learn English in a regular weekly class. T d this, a cmparisn with the secnd grup, Nrwegian university students f English was necessary. In this chapter the results f the study will be discussed and analysed. The English prficiency tests will prvide a baseline fr cmparisn, but the main interest fr discussin here will be the idim tests. The tw grups perfrmance with idims shuld prvide a view t their cmprehensin f idimatic expressins. The detailed results frm the MuseTracker sftware shuld prvide an enhanced insight int the prcessing f idims fr the grup f adlescents. First, I will discuss the cmpared results by lking at the prficiency tests and the idim accuracy. Secnd, I will discuss the results frm the MuseTracker. I will draw parallels t theries n prcessing and representatin f idims t attempt an explanatin f the results. 5.1 The English Native speakers The native speakers had a high level f crrectness fr mst f the idims we presented t them, bth fr images and written cntext. Hwever, seven idims failed t gain success in their test. These idims were T nt be playing with a full deck, T crss smene s mind, A ht ptat, Blckbuster, T put all eggs in ne basket, It takes tw t tang, Sleep with the fishes (See Appendix 3 and 4 fr their cntexts. They are marked with red asterisks). These were initially selected fr exclusin frm the results, but as we analysed the Nrwegian university students, it became clear that these idims had nt been a prblem. Nne f the idims mentined abve had a lwer crrect rate than 75% fr the Nrwegian university students, actually nly tw f them scred lwer than 90%. Fr this reasn, and the fact that the tests the natives and the Nrwegians had received were exactly the same, I decided t include these idims. They turned ut t be a bit ambiguus fr the upper secndary students, wh had prblems with sme f them, in particular T put all yur eggs in ne basket (54% crrect and lwest scre fr upper secndary, image idims), but neither f them really std ut frm the rest f the idims. There are varius explanatins frm thery that can explain this result. One culd be that the idim cntext was biased fr L2 speakers. Since the tests were all made by me and my lab partner wh are bth Nrwegians with English as L2, they culd be biased. The natives may 37

46 have had a disagreement with the cntext r the alternatives fr respnses. In the respnses they made fr 5 f these idims, they were mstly respnding with the same mistake. Tw idims were exceptins, with a spread in respnses. Anther explanatin culd be that the Nrwegian university students have a better trained metalinguistic awareness. The grup cnsisted f skilled L2 speakers, and they may have been heavily expsed t idims in the L2. This in turn sparks the familiarity effect, which wuld be a natural explanatin t gd scres. By this reasning the Nrwegian university students utperfrm English native speakers frm first year English university studies. The seven idims vary in cmpsitinality frm nrmally decmpsable and abnrmally decmpsable t semantically decmpsable idims. Thus the explanatin can nt be fund in theries n the effect f cmpsitinality as referenced by Titne and Cnnine (1994). A different explanatin culd be that the idims d nt represent widely used idims in the English language, r in that particular ppulatin f the grup. Lking at sme f the idims, like ptat, fishes, blckbuster and eggs, we culd reasnably see them as less ppular idims. Hwever, why the idim t crss smene s mind happened n this list was beynd ur understanding The English Prficiency test and Idim Accuracy The results f the prficiency tests were quite as expected. Althugh these tests alne are nt enugh t cnclude anything n the actual prficiency f the participants, they give a small indicatin f what results ne culd expect frm the tw grups. The difference in scres between the tw grups are almst the same fr the vcabulary (University: 63%, Upper Secndary: 39%) test as the grammar test (University: 82%, Upper Secndary: 56%). The scres fr the university students indicated a high level f English skills and vcabulary, accrding t the results descriptin end screen. The standard deviatin fr the upper secndary grup was ntably higher than fr the university students, indicating that the yunger grup cntained a variety f higher and lwer-level English cmprehenders. This was expected due t the nature f the tw grups. The university students were all selected frm advanced curses f English, mst f them in their third year f English studies r higher. The upper secndary grup was a class cmprised f yung learners f English as L2, nt an in-depth curse cmpared t university-level English curses. Mst f thse participants are likely t elect a different path than English studies later in their lives. The results described here was predicted t be smewhat mirrred in the idim tests. The spread 38

47 f English cmprehensin level illustrated by the standard deviatin in the idim tests shuld prvide an interesting base fr cmparisn against the results f the idim tests. Benelli et al. (2006) argued that metalinguistic ability crrelates with cgnitive develpment, metacgnitin, literacy, ral language skills and years f schling. These are all features that n dubt shuld be a great deal mre reflected in the university students than in the upper secndary students. The GEM by Levrat and Cacciari (2002) indicates that metalinguistic awareness increases with age. They cnfirmed this in a study where they had yunger and lder participants tested by asking them t create nvel expressins. The lder participants created nvel figurative expressins that were judged mre apprpriate than thse f the yunger participants (2002). As many studies n idim prcessing and cmprehensin makes a pint f dividing skilled cmprehenders frm less skilled cmprehenders, and lder frm yunger, I will make a pint f distinguishing the tw grups fr the same criteria. The upper secndary grup, cnsisting f 16 t 17 year-lds, is by nature a varied grup in terms f metalinguistic skills. As discussed earlier here, they are in mandatry classes f English, and may r may nt be particularly interested r expsed t much English in their everyday lives. The backgrund schemas cllected frm them shwed that they varied much in questins regarding hw much English they wrte, heard and spke every day. Sme f them had visited English-speaking cuntries, mst f them had nt. Due t the scpe f this study I will nt g in debth f data frm the backgrund questinnaires. The results frm the idim tests were n ne hand surprising and n the ther hand as expected. The university students perfrmed at ceiling effect in the idim test by bth stimuli. They perfrmed s well it wuld be difficult t cnclude anything n them ther than the fact that the test culd have been t easy fr them. Yet there are a few effects there that align with idim prcessing theries. Illustrated in Table 2 in the previus chapter, the results indicated that bth grups fund the written idim tasks (University students: 93% crrect, upper secndary: 61% crrect) mre difficult than the image tasks (university students: 98% crrect, upper secndary: 75% crrect). Additinally, the upper secndary grup had a much higher tendency twards a spread within the grup. Their standard deviatin values fr crrect respnses ranged abve 20 fr bth written and image idim tasks. This is strngly ppsed frm the university students, wh had standard deviatin values f 2.4 in image and 5.4 in written idim tasks. It is likely that the upper secndary grup represents a variety f cmprehenders between gd and bad. Further discussin n this tpic will cnsider why 39

48 these tests shwed such a huge distinctin between the tw grups, as well as within the upper secndary grup. Oakhill et al. (2016) fund similar effects in their study n skilled and less skilled cmprehenders. Evaluating n-line prcessing and interpretatin by measuring reading times and cmpetence they cncluded that the gd cmprehenders spent mre time reading the tasks, indicating that they activated figurative interpretatins. Their study shwed that because f the gap in metalinguistic awareness between the tw grups, the strnger grup came ahead because f their ability t infer frm the given cntext. Althugh we were nt able t measure time spent fr the university grup, we culd argue the same frm the pint f cmpetence. The cntext presented in the idim tasks was limited t a sentence r tw in the written tasks, and additinally fur alternative images in the image tasks. It is pssible that within the upper secndary grup, many participants struggled t frm an apprpriate figurative interpretatin based n this cntext. With reference t Oakhill (2016) and the GEM, the cntext given fr such idim tasks ften give the lder and mre skilled cmprehenders a bigger benefit than fr the yunger and less skilled. This culd be ne f the reasns the university students utperfrmed the yunger grup s remarkably. On the ther hand, an explanatin culd be that the cntext simply was t scarce. Nippld and Taylr (1995) emphasized that adlescents struggle with idimatic expressins withut sufficient cntext, as culd be the case fr this study. Cntext in particular is stressed in many idim prcessing accunts. Vega-Mren (2001) illustrates hw prcessing might functin in different situatins, and suggested that the infrmatin available in the cntext f an idim ften decides hw the n-line prcessing prceeds. This leads t the dimensin f familiarity and the rle it may have played in this study. With reference t Nippld s three metalinguistic strategies fr language acquisitin, the wrld experience in particular cnnects with familiarity. The lder grup shuld exceed the yunger grup nt nly by their experience with advanced English training, but als by a sheer package f metalinguistic abilities, including cultural and etymlgical knwledge. T cmpare the tw grups wuld nt nly be t assume that the lder grup hlds familiarity twards mre idims than des the yunger ne. It must be assumed als that the lder grup culd equip a set f metalinguistic abilities, making the tasks much easier. Fr example, if a university student encuntered an unfamiliar idim in the tasks, the student wuld be able t (a) infer meaning frm the cntext, (b) rule ut alternatives based n knwledge f culture r etymlgy, (c) infer meaning frm lexical items in cmbinatin with the frmer strategies. 40

49 The effects f metalinguistic awareness and familiarity were mst visible in the results f especially paque idims. Fr example, the idim Kick the bucket in the image test, shwed a very plarized result between the tw grups. The university students perfrmed a perfect scre, 100% crrect respnses. The upper secndary students achieved 69% crrect respnses, and interestingly 27% literal respnses. The literal image prtrayed an angrylking man kicking a bucket. It is reasnable t assume that this idim was unfamiliar t many f the upper secndary students. Nevertheless, the literal image shuld have been the cue fr the participants t realize that there is a figurative meaning behind the expressin. Because ur test was cnstructed with bvius literal alternatives, the participants were given a kind f a distress signal twards the figurative meaning. This example culd indicate that the metalinguistic awareness f the upper secndary participants was nt activated. In line with idim representatin and prcessing hyptheses, the yunger participants culd have been cmputing the expressins n-line fr the literal meaning. Yet, as bth the Literal first and Simultaneus hyptheses state, after the literal interpretatin has been cnsidered, it shuld be rejected fr a figurative interpretatin. It culd be the case that the participants wh respnded with literal images never apprached the idimatic key in these expressins. In that case, they must have fund the literal image t be relevant t the intended meaning f the idim s cntext. Oakhill (2016) suggested that cntext influences idim cmprehensin in tw ways: (a) the reader cnstructs a cherent representatin f the meaning f the task, text and image and realizes that the literal interpretatin must be inapprpriate, and (b) the reader attempts t derive a figurative meaning frm the cntext while prcessing, in line with Cnfiguratin hypthesis and Phrase-Induced Plysemy (PIP). This crrelates with Levrat, Nesi and Cacciari, wh suggested several reasns why less skilled cmprehenders wuld respnd with literal alternatives despite idimatic cntexts. The three pssible reasns were (a) their inability t suppress the literal interpretatin frm the wrds in the expressin, (b) reduced ability t infer frm cntext, and (c) limited ability t prcess and infer an idimatic meaning (2004). The last issue t be discussed frm the accuracy tpic, is the difference between image and written tasks. As illustrated in table 2, it is clear that image tasks were much easier than the written tasks. Especially s fr the upper secndary grup wh sank 14,5 pints in crrect respnses frm image t written. Even mre remarkable were their tendency t chse distractrs. Actually they did s nly in 11% f the tasks fr the image, but a vast 29% f the time in written tasks. It seems they were easily distracted and attracted by the distractr 41

50 images, which we cnstructed t be smewhat arbitrary and smewhat related t the cntext. This aligns with Cper s (1999) study, in which subjects tended t guess frm cntext when faced with unfamiliar idim tasks MuseTracker data Reactin Times The results shwed a tendency twards higher RT fr incrrect respnses than crrect respnses (see Table 3 in Results). The mean (crrect: 2068ms, incrrect: 4086ms) and standard deviatin (crrect: 2539ms, incrrect: 4707ms) was twice the size in incrrect respnses than fr the crrect respnses. The tendency was illustrated in the Results chapter by the 6 mst plarized idims in terms f reactin times. The three idims with highest RT, Bent ut f shape, Cst an arm and a leg, Put all eggs in ne basket, are amng the idims with mst incrrect respnses. The three ppsing idims, Piece f cake, Hit the hay and Kill tw birds with ne stne, with very high success rate in the MuseTracker test, had the lwest mean RT s. The reasn fr this is likely t reside in the nature f idim prcessing and representatin. Assuming the tasks and the cntext f the idims was gd and fair fr the upper secndary participants, the high RT may indicate an encunter with fully r partially unfamiliar idims. Fllwing this case, and Vega-Mren s accunt (2001) n prcessing unfamiliar idims, the different RT results may be explained. Encuntering an unfamiliar idim, the participants decde the cncepts in the cnstituents f the string, and prcesses the cntext infrmatin. In the given task, lexical infrmatin might give clues t a literal interpretatin. At sme pint the participant realizes the deviatin between the literal meaning and the intended meaning, triggering the idimatic key. This shuld be easier in bilgical idims than cultural and instructive idims, as referenced frm Vulchanva et al. (2011). Regarding the 6 plarized idims, hwever, the three highest RT idims leaned twards bilgical and cultural typlgies (fr example, Get bent ut f shape and Cst and arm and a leg shuld fall int ne f the tw categries). S were the lw RT idims (fr example It rains cats and dgs and Bite ver mre than yu can chew ). Althugh we did nt take measures t prvide ur tests with secure typlgies due t the scpe f this study, s this is incnclusive. Althugh studies frm the chapter n theretical framewrk reprt lnger prcessing times fr participants wh succeed in interpreting figurative meanings (Gernsbacher & Glucksberg, 1984; Gibbs, 1980; Swinney & Cutler, 1979; Titne & Cnnine, 1994, 1999), we must assume that the RT s rather shw a result f certainty and familiarity. With this reasning, the 42

51 participants with high RT were either familiar with the idimatic expressins in the test, r equipped with a mre develped metalinguistic awareness than the participants with lwer RT s. Bth the frmer and the latter ability wuld f curse rely n the assumptin that the cntext and the alternatives f the tasks were representative and understandable fr the participants Respnse Certainty The respnse certainty data was the mst difficult t cllect, and prbably the mst rewarding in relatin t the tpic f this study. A quantified measure fr the certainty f the participants hand-mvements was needed in rder t analyse their trail f thught as they perfrmed the tasks. In ther wrds, this data shuld reflect their n-line prcessing as they make a decisin based n interpretatin f the idim and its cntext. First I will pint t the general trend f the hand-mvements between crrect and incrrect respnses. After that I will attempt t explain the trend with thery n idim prcessing. As illustrated in Figures 7-10 in the Results chapter, the general tendency fr the crrect respnses is that f a straight line. The AUC (Figure 9) shws a rbust mean with little spread. The MD (Figure 7) is a bit mre skewed. The tendency is still clearly shwing trajectries ging with sme certainty twards the crrect target, but a few f the idims have deviatins frm the mean. This indicates that ther targets than the crrect have been cnsidered in the prcess. There is a trend that the idims with mst crrect respnses verall have mre certain trajectries, just like with the RT. See fr example hay ( hit the hay ) and chew (bite ver mre than yu can chew) in Figure 7. Especially dubtful idims in crrect respnses were Put all eggs in ne basket and tie the knt. The figures fr incrrect respnses, Figure 8 and 10, tell a very different stry. Again, the AUC figure des nt really shw very much tendency away frm the mean. As can be seen in Figure 10, sme f the idims prvide very large spreads, like rain ( it s raining cats and dgs ), but this is misleading because there are nly 4 incrrect respnses fr that idims. One f these respnses made, ver a 33 secnd time curse, a very errneus trajectry. The ther three respnses did in fact have quite nrmal trajectries, s nthing can be cncluded frm that result. Idims that did, hwever, prvide interesting results by AUC were bat ( t be in the same bat ), fish ( sleep with the fishes ) and hear ( brken heart ). The MD statistics fr incrrect respnses, as illustrated in Figure 8, shw a clear trend f attractin twards unselected targets. As expected frm incrrect respnses, the degree f certainty shuld be lwer than fr crrect respnses. 43

52 Why exactly sme idims create mre uncertainty than thers, is interesting with regards t prcessing and representatin f idims. Nevertheless, the items that prduce very deviant means and spreads frm the straight line pattern, tend t cnsist f few respnses. Take fr example the fur mst deviant idims fr incrrect respnses by MD, bat ( be in the same bat ), heart ( brken heart ), lse ( t lse yur head ) and ptat ( cuch ptat ). The MD figures illustrate a strng deviatin, but the respnses in sum amunt t 22. Mst f these idims generated high scres fr crrectness, except frm bent, which generated 10 errrs. Lking back at the fur highest RT idims, f which bent is a member, is interesting because they have generated 32 errrs ur f 100. The trajectries fr these idims shw great uncertainty, illustrated in Figure 11 belw. Figure 11. Fur selected idim trajectries. It is clear that the trajectries in the right-hand bx are headed anywhere but in a straight line twards a single target. This indicates that the n-line prcessing culd be measuring the different alternatives up against each ther, befre settling dubtfully upn an incrrect respnse. Of interest, the nly idims f the list f unacceptable idims as judged by English native speakers t appear n these deviatins are eggs and ptat. Apparently, the 7 idims judged as unacceptable and less cmprehensible than the rest, turned ut t be less unacceptable and incmprehensible fr the upper secndary students (this is cmpared t the rest f the results frm the upper secndary grup). The results shuld reflect and shw an ccurrence f hw prcessing and representatin f idims in adlescents wrks. It fits Vega- Mren s (2001) descriptin f prcessing well, in particular the prcessing f unfamiliar idims since it is assumed in this study that the incrrect results prbably highlight the familiarity effect mre than anything. 44

53 5.3 General Discussin The results f this study reflect the GEM well. Cmparing the tw grups fr accuracy gives an insight t the discrepancy between metalinguistic skills. The cmpsitin f the tw grups highlight the factrs fr metalinguistic awareness well. The university students represent a high level, with their backgrund frm advanced English L2. Perfrming at ceiling in the idim accuracy tests, and having gd scres in English prficiency fr L2 speakers, they represent a grup f L2 idim cmprehenders which is nt much subjected t research in prcessing and representatin. In fact, they are lcated beynd the GEM, which states that phase 5 and nwards is a phase f grwing metalinguistic awareness. The upper secndary grup represent a standard VG1 class with much less English L2 backgrund. Theries n idim cmprehensin emphasize the imprtance f L2 expsure, wrld knwledge, cultural knwledge and general English prficiency in rder t build a strnger metalinguistic awareness. In line with the GEM, the upper secndary grup reflected an adlescence grup smewhere between phase 4 and 5 as described by Levrat and Cacciari (2002). The certainty f respnse data illustrated the n-line prcessing f the upper secndary grups. The tendency f the hand-mvements fr this grup was clear in the directin f uncertainty when they respnded incrrectly. This culd explain their higher scre f incrrectness in tw pssibilities: (a) the cntext and alternatives prvided were insufficient r misleading fr their metalinguistic awareness t trigger and prduce a crrect respnse, which in turn resulted in very cnfused hand-mvements. Or (b) many f the idims were unfamiliar t them, leading t a prcess in which they attempted t interpret the idims nline and fr many f them this prcess lead t either an incrrect respnse in favur f an image which reflected the cntext but nt the intended meaning, r a literal interpretatin. The chice f literal interpretatins was a much larger trend fr the upper secndary than fr the university students, supprting the thery f the GEM and metalinguistic awareness theries. 45

54

55 6. Cnclusin The aim f this study was t test and investigate Nrwegian upper secndary students fr cmprehensin and prcessing f idims. Twenty-six Nrwegian upper secndary students frm a VG1-class, plus twenty-six Nrwegian university students f English were tested. Bth grups cmpleted tw sets f English prficiency tests, ne fr vcabulary and ne fr grammar. Then they cmpleted tw different idim tests, ne with written alternatives and ne with image alternatives. The university students cmpleted bth by a questinnaire frmat, while the upper secndary cmpleted the written test by a questinnaire frmat and the image test by a MuseTracker sftware frmat. Furthermre, ninety-nine native speakers f English frm a university in England were tested fr the same idim tests by questinnaire, which was used t judge the validity f the tests. A cmparisn was made between the tw main grups, with the English prficiency tests as a baseline, and the idim tests. The tests shws that the university student grup utperfrmed the upper secndary grup heavily in bth prficiency and idim cmprehensin. An analysis was made n this based n thery f idim prcessing and representatin. Several idims in particular were picked ut and analysed in an attempt t explain the results. Furthermre, the upper secndary grup s results in the MuseTracker sftware was analysed in rder t explain hw they prcessed the idims n-line. In line with theries f idim prcessing and metalinguistic awareness, the university students utperfrmed the upper secndary students due t a backgrund difference and age gap. Metalinguistic abilities were assumed t be instrumental in separating the gd frm the bad perfrmers in these tests, and this was smething the upper secndary students lacked in. Factrs like cntext and familiarity was assumed t be hindrances t the upper secndary students, wh tended t respnd with literal alternatives and distractrs in many cases. This was nt the case fr the university students, wh perfrmed at ceiling in the idim tests. In fact, the Nrwegian students utperfrmed even the English native speakers, which allwed this study t verrule the idim tasks judged as unacceptable by the native speakers. The results shws that metalinguistic abilities in Nrwegians with English as L2 has a big gap between upper secndary and university level. Age, years f schling and expsure t the L2 is imprtant t understand figurative expressins in different situatins. Such expressins amunt t a significant part f all languages, and cme in different sizes and variances, and range frm very transparent t cmpletely paque in cmpsitinality. This makes them a 47

56 difficult bulk f language t vercme. Cmprehenders have t either acquire the expressins t the lng-term memry r train their metalinguistic abilities t a high level in rder t understand them. 6.1 Questins fr further research The very limited amunt f research cnducted n English L2 speakers in Nrway leaves the subject pen. The findings f this study can nly hint at an area f interest. What kind f idims des this ppulatin cmprehend r acquire easiest? Tests cntrlled fr typlgies culd reveal a difference between Nrwegians and native speakers f English, r even between gd and bad Nrwegian cmprehenders f L2 English. Furthermre, this study culd nt cntrl fr familiarity. A study cntrlled fr familiarity culd, fr example, translate idims frm anther language int an understandable Nrwegian translatin, and prceed t have Nrwegian speakers f L2 English judge them. Many studies have dne this in ther languages, because it prvides the test with nvel idimatic expressins which is sufficiently cntrlled fr familiarity. 48

57 References Altmann, G. T. M. (1997). The ascent f Babel : an explratin f language, mind, and understanding. Oxfrd: Oxfrd University Press. Benelli, B., Belacchi, C., Gini, G., & Lucangeli, D. (2006). 'T Define Means t Say What Yu Knw abut Things': The Develpment f Definitinal Skills as Metalinguistic Acquisitin. Jurnal f Child Language, 33(1), di: /s Bbrw, S., & Bell, S. (1973). n catching n t idimatic expressins. Memry and Cgnitin(1), Cacciari, C., & Glucksberg, S. (1991). Understanding idimatic expressins: the cntributin f wrd meanings. In G. B. Simpsn (Ed.), Undestanding Wrd and Sentence (pp ). Amsterdam: Elsevier. Cacciari, C., & Tabssi, P. (1988). The Cmprehensin f Idims. Jurnal f Memry and Language, 27(6), di: / x(88) Cambridge University Press. (2006). Grammar test, C2 level. Retrieved Nvember 2015, frm Chmsky, N. (1965). Aspects f the Thery f Syntax. Cambridge: Cambridge: MIT Press. Cllins Cbuild Dictinary f Idims. (1995). (2 ed.). Glasgw: HarperCllins Publishers. Cper, T. C. (1999). Prcessing f Idims by L2 Learners f English. TESOL Quarterly, 33(2), Ellis, N. C. (2002). Frequency Effects in Language Prcessing: A Review with Implicatins fr Theries f Implicit and Explicit Language Acquisitin. Studies in Secnd Language Acquisitin, 24(2), di: /s Freeman, J., & Ambady, N. (2010). MuseTracker: Sftware fr studying real-time mental prcessing using a cmputer muse-tracking methd. Behavir Research Methds, 42(1), Freeman, J. B., Dale, R., & Farmer, T. A. (2011). Hand in Mtin Reveals Mind in Mtin. Frntiers in Psychlgy, 2. di: /fpsyg Gass, S. M., & Selinker, L. (2001). Secnd language acquisitin : an intrductry curse (2nd ed. ed.). Mahwah, N.J: LEA. Gernsbacher, M. A., & Glucksberg, S. (1984). Reslving 20 years f incnsistent interactins between lexical familiarity and rthgraphy, cncreteness, and plysemy. Jurnal f 49

58 Experimental Psychlgy: General, 113(2), di: / Gibbs, R. (1980). Spilling the beans n understanding and memry fr idims in cnversatin. Memry and Cgnitin, 8, Gibbs, R. W., Jr., Nayak, N. P., & Cutting, C. (1989). Hw t Kick the Bucket and Nt Decmpse: Analyzability and Idim Prcessing. Jurnal f Memry and Language, 28(5), di: / x(89) Glucksberg, S., Brwn, M., & McGlne, M. S. (1993). Cnceptual metaphrs are nt autmatically accessed during idim cmprehensin. Memry and Cgnitin, 21(5), Jackendff, R. (2002). What's in the lexicn? In S. G. Ntebm, F. Weerman, & F. Wijnen (Eds.), Strage and cmputatin in the language faculty (Vl. vl. 30, pp ). Drdrecht: Kluwer Academic. Levrat, M. C., & Cacciari, C. (1992). Children's Cmprehensin and Prductin f Idims: The Rle f Cntext and Familiarity. Jurnal f Child Language, 19(2), di: /s Levrat, M. C., & Cacciari, C. (1995). The Effects f Different Tasks n the Cmprehensin and Prductin f Idims in Children. Jurnal f Experimental Child Psychlgy, 60(2), di: /jecp Levrat, M. C., & Cacciari, C. (2002). The Creatin f New Figurative Expressins: Psychlinguistic Evidence in Italian Children, Adlescents and Adults. Jurnal f Child Language, 29(1), di: /s Levrat, M. C., Nesi, B., & Cacciari, C. (2004). Reading Cmprehensin and Understanding Idimatic Expressins: A Develpmental Study. Brain and Language, 91(3), di: /j.bandl Marslen-Wilsn, W. D. (1987). Functinal Parallelism in Spken Wrd-Recgnitin. Cgnitin: Internatinal Jurnal f Cgnitive Science, 25(1-2), di: / (87) Nippld. M. A. (2006). Language Develpment in Schl-Age children, Adlescents, and Adults. In K. Brwn (Ed.), Encyclpedia f Language and Linguistics (Vl. 6, pp ). Nippld, M. A., & Duthie, J. (2003). Mental imagery and idim cmprehensin: A cmparisn f schl-age children and adults. Jurnal f Speech, Language, and Hearing Research, 46(4),

59 Nippld, M. A., Taylr, N. (1995). Idim understanding in yuth: further examinatin f familiarity and transparency. Jurnal f Speech and Hearing Research(38), Nippld, M. A., & Rudzinski, M. (1993). Familiarity and transparency in idim explanatin: A develpment study f children and adlescents. Jurnal f Speech and Hearing Research, 36, Nunberg, G. (1978). The Pragmatics f Reference: Indiana University Linguistic Club. Oakhill, J., Cain, K., & Nesi, B. (2016). Understanding f Idimatic Expressins in Cntext in Skilled and Less Skilled Cmprehenders: Online Prcessing and Interpretatin. Scientific Studies f Reading, 20(2), di: / Odlin, T. (2003). Crss-Linguistic Influence. In C. Dughty & M. Lng (Eds.), The Handbk f Bilingualism (pp ): Blackwell Publishing Ltd. Pavlenk, A. (1999). New appraches t cncepts in bilingual memry. Bilingualism: Language and Cgnitin(2), Plli, H. R. (1977). Psychlgy and the petics f grwth : figurative language in psychlgy, psychtherapy, and educatin. Hillsdale, NJ: L. Erlbaum Assciates. Styanva, M. K. (2013). Near-native speakers' recgnitin and prductin f idims in L2. (PHD), Auburn University, Mntgmery. Swinney, D. A., & Cutler, A. (1979). The access and prcessing f idimatic expressins. Jurnal f Verbal Learning and Verbal Behavir(18), Titne, D. A., & Cnnine, C. M. (1994). Descriptive Nrms fr 171 Idimatic Expressins: Familiarity, Cmpsitinality, Predictability, and Literality. Metaphr and Symblic Activity, 9(4), di: /s ms0904_1 Titne, D. A., & Cnnine, C. M. (1999). On the Cmpsitinal and Nncmpsitinal Nature f Idimatic Expressins. Jurnal f Pragmatics: An Interdisciplinary Jurnal f Language Studies, 31(12), di: /s (99) Ugent.be (2015). Vcabulary test. Retrieved Nvember, 2015, frm Vega-Mren, R. E. (2001). Representing and prcessing idims. Verbal learning and Verbal Behaviur, 18, Vulchanva, M., Vulchanv, V., & Stankva, M. (2011). Idim cmprehensin in the first language: A develpmental study. Vig Internatinal Jurnal f Applied Linguistics, 8(1), Zyzik, E. (2010). Sin pels en la lengua: La adquisición de mdisms en una clase de españl cm lengua extranjera. Hispania, 93(3),

60

61 Appendices Appendix 1: Cnsent Frm Frespørsel m deltakelse i frskningsprsjektet Frskningsprsjekt på Engelsk språk g rdtak Bakgrunn g frmål Denne studien er en del av et mastergradsprsjekt der vi er interesserte i å avdekke g kartlegge deres evner innen engelsk sm andrespråk g da spesielt med fkus på rdtaksfrståelsen. Vi studerer på NTNU ved fremmedspråklig institusjn på Dragvll, g samarbeider med veiledere Mila Vulchanva g Giuse Baggi m dette prsjektet. Dere er valgt ut frdi vi ønsker en genuin gruppe med nrske 16/17-åringer sm har engelsk sm andrespråk eller eventuelt tredje. Hva innebærer deltakelse i studien? Om dere samtykker til å delta vil dere bidra med veldig gd infrmasjn sm vil kunne gi ss mer kunnskap m engelsk språk i Nrge. Dette kan igjen bidra til å gjøre f. eks. utdanning g pplæring bedre på sikt. Dette studiet vil ikke ha ne å si fr din karakter eller deltakelse i sklen ellers. Det bes m at alle svarer ærlig g ikke prøver å gå gjennm nen av testene mer enn en gang, da dette vil svekke våre resultater. Selve prsjektet vil trenge deres aktive deltakelse i cirka 1-2 skletimer. Vi vil begynne med t «tester» der dere får svare på nen spørsmål m engelsk grammatikk g rdtak. I grammatikk-delen vil dere få spørsmål m dere har kjennskap til en rekke rd, mens i rdtak-delen vil dere få presentert 30 rdtak i tur, med fire alternative svar til hvert av dem. Den andre delen har vi med ss egne laptper, g dere vil få en undersøkelse bestående av 30 rdtak med fire mulige alternative svar sm vi har tegnet. Med dette får vi samlet all den infrmasjnen vi trenger, g det vil bli analysert på et senere tidspunkt, selvfølgelig helt annymisert. Hva skjer med infrmasjnen m deg? Alle persnpplysninger vil bli behandlet knfidensielt. Dere vil få eller har fått utdelt nen kder av lærere sm vil bli brukt på testen. Listen med navn g kder har lærer kntrll på under studiet, men etter dette vil det bli låst inne i en skuff av vår veilder i et mråde på NTNU der man trenger et spesielt nøkkelkrt fr å få tilgang. Deres persnpplysninger blir altså ikke tilgjengelig fr nen, g i selve studiet er dere kun en gruppe med kder g derfr annyme. Fr flk sm leser studiet eller funnene våre vil det ikke være på nen måte mulig å gjenkjenne hvem sm har deltatt. Den vil prsjektet slutte, g infrmasjnen vil frbli innlåst Frivillig deltakelse Det er frivillig å delta i studien, g du kan når sm helst trekke ditt samtykke uten å ppgi nen grunn. Dersm du trekker deg, vil alle pplysninger m deg bli annymisert. Studien er meldt til Persnvernmbudet fr frskning, Nrsk samfunnsvitenskapelig datatjeneste AS. 53

62 Samtykke til deltakelse i studien Jeg har mttatt infrmasjn m studien, g er villig til å delta (Signert av prsjektdeltaker, dat) 54

63 Appendix 2: Backgrund Questinnaire Bakgrunnsinfrmasjn fr frskningsprsjekt m andrespråksfrståelse Tusen takk fr at du har sagt ja til å delta i vårt frskningsprsjekt m andrespråksfrståelse. I dette skjemaet ber vi m bakgrunnsinfrmasjn sm er nødvendig fr at resultatene fra undersøkelsen skal kunne brukes. Infrmasjnen sm du ppgir vil bli behandlet uten direkte gjenkjennende pplysninger. En kde knytter deg til dine pplysninger gjennm en deltakterliste. Det er kun autrisert persnell knyttet til prsjektet sm har adgang til deltakerlisten g sm kan finne tilbake til infrmasjnen. Del B, C g D av dette skjemaet vil bare ppbevares med kden. All infrmasjn vil bli annymisert ved prsjektslutt. Det vil ikke være mulig å identifisere deg i resultatene av studien når disse publiseres. Vi ber deg legge merke til at skjemaet har ttalt 6 sider. Vegard Bergh & Anders Hauge Aurland Studenter ved lektrutdanningen, ved Institutt fr språk, Engelsk, NTNU Del A: Persnlig infrmasjn Studieretning g trinn: Fødselsår Kjønn Kvinne Mann Bstedskmmune 55

64 Deltakerkde: Del B: Språklig bakgrunn Mrsmål Er nrsk mrsmålet ditt? Ja Nei Hvis ja, har du andre mrsmål i tillegg? ja Nei Hvis ja, hvilke(t) språk? Hvilket språk bruker dere hjemme? Hvr fte leser du tekst skrevet på nrsk? Hver dag Flere ganger i uka Et par ganger i uka Av g til Aldri Hvr fte skriver du tekst på nrsk? Hver dag Flere ganger i uka Et par ganger i uka Av g til Aldri Engelsk g andre fremmedspråk I engelsk, hvrdan vurderer du ferdighetene dine på hvert av disse mrådene? Lesing Skriving Snakke Lytte Grunnleggende Middels Avansert Flytende Ttalt Har du bdd i, eller hatt lengre pphld i, et land hvr engelsk er hvedspråk? 56

65 Ja Nei Hvis ja, hvr lenge varte pphldet/pphldene? Har du vært på krtere (under 14 dager) reise i et land hvr engelsk er hvedspråk? Ja Nei Har du bdd i, eller hatt lengre pphld i, et land hvr annet enn engelsk er hvedspråk? Ja Nei Hvis ja, hvr var det, g hvr lenge varte pphldet/pphldene? Hvilke språk kan du utver mrsmålet ditt g engelsk? Språk Tysk Fransk Spansk -Angi språk Nivå Grunnleggende Middels Avansert Flytende -Angi språk -Angi språk 57

66 Hvr fte leser du tekster på engelsk? Hver dag Flere ganger i uka Et par ganger i uka Av g til Aldri Hvr fte skriver du tekster på engelsk? Hver dag Flere ganger i uka Et par ganger i uka Av g til Aldri Hvr fte lytter du til/hører du engelsk? Hver dag Flere ganger i uka Et par ganger i uka Av g til Aldri Hvr fte ser du på engelskspråklige serier/filmer? Hver dag Flere ganger i uka Et par ganger i uka Av g til Aldri Når du ser på engelskspråklige filmer, hvilken av disse alternativene bruker du mest? Undertekst på nrsk Undertekst på engelsk Ingen undertekst Hvr fte ser du på engelskspråklige tegneseriefilmer/serier? Hver dag Flere ganger i uka Et par ganger i uka Av g til Aldri Hvr fte spiller du engelskspråklige data/tv-spill? Hver dag Flere ganger i uka Et par ganger i uka Av g til Aldri Hvilke type spill spiller du? Hvr mange timer cirka per dag? Hvr mye TV ser du på hver dag? 7 timer eller mer 5-6 timer 3-4 timer 1-2 timer aldri eller nesten aldri Del C: Andre faktrer i språklæring Har du, eller har du hatt, prblemer med synet utver nrmal brillebruk? Ja Nei Har du, eller har du hatt, prblemer med hørselen? Ja Nei 58

67 Har du, eller har du hatt, språkvansker av ne slag (spesifikke språkvansker, lese- /lærevansker eller lignende)? Ja Nei Har du, eller har du hatt, andre diagnser sm kan tenkes å påvirke språklæring (ADHD, autisme eller lignende)? Ja Nei Er du venstrehendt? Ja Nei Del D: Vkabulartest g grammatikktest Resultat vkabulartest: Resultat grammatikktest: 59

68 Appendix 3: Idim tasks with written cntext English idims frm 1 Chse the crrect idim frm the 4 alternatives in each questin What is the meaning f the expressin blessing in disguise? I assumed that when my new rmmate mved in, all hell wuld break lse. Turns ut she instead was a blessing in disguise. Smene wh's up t n gd A bad start means a bad ending A rmmate wh hides his r her qualities A misfrtune which eventually turns ut psitive What is the meaning f the expressin rll up ne s sleeves? As a natin, it is time fr us t rll up ur sleeves and give hmeless a hme and their dignity back. We simply cannt stand back and passively watch this prblem grw ut f cntrl. T prepare fr hard wrk T call fr the big bucks t rll in T use less mney n urselves in rder t give sme mre t the less frtunate T be cheap What is the meaning f the expressin Icing n the cake? I knew that my parents wuld cme t the graduatin ceremny, but that my grandparents wuld cme was the icing n the cake. An extra enhancement A surprise Mre than necessary An intimidating thught What is the meaning f the expressin mre r less? She had mre r less decided t get a new jb. She needed sme new challenges. Whleheartedly T dream f bigger things Slwly but certainly T sme extent What is the meaning f the expressin be game? Lauren was psting an invite t an event n Facebk asking: wh is game?. T be willing t play a game T be excited abut an upcming event T be happy T be agreeable t participate in smething 60

69 What is the meaning f the expressin take charge? It was cmplete chas after the accident. Peple were either running arund creating mre panic r just standing there watching. Luckily, a nurse arrived and tk charge. T be respnsible T make smene pay fr smething T take cntrl ver smething T relad ne's batteries What is the meaning f the expressin rule f thumb? If yu receive an frm yur bank asking fr yur passwrd, a gd rule f thumb is simply t delete it. Yur bank wuld nt ask fr such infrmatin. An answer n hw t deal with prblems A general principle develped by experience T act cautiusly in rder t prtect sensitive infrmatin and yur cmputer T measure smething using yur thumb as a ruler What is the meaning f the expressin put wl ver peple s eyes? After reading Gerge Orwell s nvel 1984 I kept fearing ur pliticians were putting wl ver ur eyes. I really hpe they re nt playing us arund. T impse ignrance n peple by hiding the truth behind less imprtant matters T play games with ther pliticians T pretend t knw abut nvels that ne haven't actually read T advertise fr clthes n TV and radi What is the meaning f the expressin nt playing with a full deck? * That by is definitely nt playing with a full deck, He s desperately trying t lick his elbw althugh he clearly wn t make it. Smene wh is nt especially clever T miss a few pieces in a puzzle T try really hard Smene wh wants t accmplish smething impssible What is the meaning f the expressin turn the tables? It s time t turn the tables she whispered. She had planned this fr weeks. This time he wuld be the victim! T change a situatin s that smene's psitin is the ppsite f what it was T plan revenge T turn all the tables yu see in rder t change a situatin T redecrate What is the meaning f the expressin crss smene s mind? * It crssed my mind when I read the article abut the fireman wh rescued five peple my purpse in life is t be a fireman. T suddenly think f smething T feel a divine inspiratin and calling frm abve 61

70 T wish yu were brave T think abut smething fr a shrt time What is the meaning f the expressin cuch ptat? * Yesterday I fund six steps fr hw t stp being a cuch ptat n Wikihw. The first step was t unplug the televisin. A persn wh has t eat chips while watching TV A persn wh is unemplyed and has nthing t d A persn wh spends much time sitting r lying dwn, usually watching TV A persn wh likes t eat dinner while watching TV What is the meaning f the expressin the ball is in yur curt? The plice fficer wh stpped Greg fr speeding in traffic tld him that the speed was dangerusly high and that it culd ve had grave cnsequences. S what d yu have t say in yur defence? The ball is entirely in yur curt, Mr. It's yur turn t act r speak The car was in the wrng curt at the wrng time Yu shuld start aplgising Yu will meet me in curt What is the meaning f the expressin back t the drawing bard? After a disastrus game f ftball, the cuch called in the team and uttered Well, bys, back t the drawing bard it is! T draw ftball in detail T perfrm an activity exactly as the drawings tell yu T retire frm activity t start drawing T step back and make new plans What is the meaning f the expressin dn t quit yur day jb? Paul insisted that his paintings were the best thing he d ever dne. His friends all laughed and said Dn t quit yur day jb, br! An indicatin that yu're nt very gd at what yu d besides yur prfessin A pleasant cmpliment given t artists wh shw prmise T lse a jb because f a hrrible mistake Smething yu say when friends laugh tgether What is the meaning f the expressin add insult t injury? The bss in ur cmpany asked everyne t humbly accept a pay cut t make up fr the bad ecnmy. We felt that this was nly adding insult t injury. T find a new slutin T disrespect smene wh is injured and thus cannt respnd T act in a way that makes a bad situatin wrse T be very rude What is the meaning f the expressin actins speak luder than wrds? 62

71 The teacher was pleased when Jhanna gave him the mney he had been missing fr days. I can t really explain this, she says. Actins speak luder than wrds the teacher respnded. That actins always are better than talking T return things withut hesitating T talk less That smetimse the act f ding smething can explain mre than anything yu culd say What is the meaning f the expressin penny fr yur thughts? Mary seemed very thughtful ne mrning. Jhn said Ha! Penny fr yur thughts, Mary!. A trick ne can play t get mney frm unaware friends T ask fr a quick lan A way t have smene ask yu a persnal questin T wnder what is n smene's mind What is the meaning f the expressin ht ptat? * The plitical parties seemed t avid the immigratin tpic befre the electin. My friend said that it certainly is this electin s ht ptat. A ptat s ht that it will shuld be put in the refridgeratr A subject which is cnsidered harmful t neself and thus gets ignred by everyne A subject that causes much fear Smething that can burn yur hands What is the meaning f the expressin beat abut the bush? Let s nt beat abut the bush the design was rejected. The emplyer said it was hrrible. We simply have t start all ver again. T run away int the bush in rder t avid facing challenges T avid talking abut a difficult subject because yu are wrried abut upsetting the persn yu are talking t T give up T get angry at neself when ne has dne smething that des nt meet the expectatin f thers What is the meaning f the expressin fit as a fiddle? I was shcked when I gt the news. I felt as fit as a fiddle until the dctr shwed me the negative results. T be in very gd health T feel secure T be psitive T be strng What is the meaning f the expressin piece f cake? They all thught the match wuld be a piece f cake but they were wrng. The ther team was faster. T be satisfactry 63

72 The best part f the cake t eat The reward yu get fr winning Smething which is easy What is the meaning f the expressin at ne s fingertips? She has the bk at her fingertips. If yu wnder where t find the infrmatin she wuld be the ne t ask. T have smething in yur hand A few cins that yu grab between yur fingertips T have quick access T be well prepared and in cntrl What is the meaning f the expressin at the tp f ne s lungs? I wke up this mrning t my neighbur singing My heart will g n at the tp f his lungs. T breathe in a cntrlled manner As ludly as ne's vice will allw T have a wish t anny smene As ut f tune as pssible What is the meaning f the expressin n a rll? I had been wrking fr nine hurs but I culdn t stp. I was n a rll. T eat spring rlls while wrking T be in the midst f a series f successes T be a wrk-a-hlic T live up t the expectatins in rder t stay n the cmpany's payrll What is the meaning f the expressin take advantage f? In tday s newspaper I read an article with the fllwing headline: Prpane cmpany takes advantage f custmers. It turned ut that I was ne f them. I checked the last mnth s bill they had vercharged me t! T imprve a situatin T d what seems best fr the cmpany T explit T disrespect What is the meaning f the expressin blckbuster? * Mark and steve were fighting abut which blckbuster t bring hme and watch: Spectre r The Dark Knight Rises? Mvies that will blw the viewer's mind A new man n the blck Explsives and gunshts in huge numbers Smething prduced t achieve high ppularity and huge sales What is the meaning f the expressin by the bk? 64

73 He was quite certain that he wuld nt be in truble when the plice came. He had dne everything by the bk. When a mvie is based n a bk T d exactly what yur emplyer tells yu t d T fllw the rules exactly T d all yur wrk by the bk shelf What is the meaning f the expressin slippery slpe? The questin f assisted death is a slippery slpe in terms f mral and legal cnsideratins. A path which is likely t lead t failure r serius truble An icy hill A persnal questin A questin that requires a decisin What is the meaning f the expressin "nt my cup f tea"? Sme peple lve skiing, but it's nt my cup f tea. I prefer swimming. Smething yu enjy r like Smething yu dn't like A brrwed cup used fr drinking tea A type f tea yu dn't like What is the meaning f the expressin "sweet tth" Every time we watch a mvie with Hans, he has t bring all the candy. I dn't understand hw he can eat all that sugar? T have diabetes T have a gld tth T desire t eat sweet fds T indicate in a humristic way that ne has a prblem with ne's health 65

74 Appendix 4: Idim tasks with image alternatives Chse the picture that shws the meaning f "t flip smene the bird" Flip the bird Tim was angry with Sara. He flipped her the bird. A B C D 66

75 Chse the picture that shws the meaning f "t have a brken heart" Brken heart Tim was sad. He had a brken heart. A B C D 67

76 Chse the picture that shws the meaning f "t be caught red handed" Red handed Tim was up t n gd. Jim caught him red handed. A B C D 68

77 Chse the picture that shws the meaning f "t have cld feet" Cld feet Jim was starting t regret his decisin. He was having cld feet. A B C D 69

78 Chse the picture that shws the meaning f "t cst an arm and a leg" Cst an arm and a leg Tim really wanted the apple. It cst him an arm and a leg. A B C D 70

79 Chse the picture that shws the meaning f "t be a cuch ptat" Cuch ptat * Sara likes t relax. Sme wuld call her a cuch ptat. A B C D 71

80 Chse the picture that shws the meaning f "t put all yur eggs in ne basket" Put all yur eggs in ne basket * Jim made a pr decisin. He suldn't have put all his egs in ne basket. A B C D 72

81 Chse the picture that shws the meaning f "t feel a bit under the weather" Feeling a bit under the weather Tim was having a rather pr day. He was feeling a bit under the weather. A B C D 73

82 Chse the picture that shws the meaning f "t get bent ut f shape" Get bent ut f shape Jim was in a bad md. Tim tld him t nt get bent ut f shape. A B C D 74

83 Chse the picture that shws the meaning f "t hit the hay" Hit the hay Tim was feeling tired. He decided t hit the hay. A B C D 75

84 Chse the picture that shws the meaning f "t give smene a hand" Give smene a hand Sara saw Jim was in truble. She decided t give him a hand. A B C D 76

85 Chse the picture that shws the meaning f "t have green fingers" Green fingers Tim is gd with plants. He has green fingers. A B C D 77

86 Chse the picture that shws the meaning f "t bite ff mre than yu can chew" Bite ff mre than yu can chew Sara had made a mistake. She had bitten ff mre than she culd chew. A B C D 78

87 Chse the picture that shws the meaning f "t kill tw birds with ne stne" Kill tw birds with ne stne Tim was in a hurry. He decided t kill t birds with ne stne. A B C D 79

88 Chse the picture that shws the meaning f "t kick the bucket" Kick the bucket Tim had an accident. He kicked the bucket. A B C D 80

89 Chse the picture that shws the meaning f "it takes tw t tang" It takes tw t tang * Jim culdn't d it alne. He quickly understd it takes tw t tang. A B C D 81

90 Chse the picture that shws the meaning f "t be in the same bat" Be in the same bat Jim and Tim bth made pr chices. Nw they are in the same bat. A B C D 82

91 Chse the picture that shws the meaning f "t hit the nail n the head" Hit the nail n the head Tim is very clever. He ften hits the nail n the head. A B C D 83

92 Chse the picture that shws the meaning f "t tie the knt" Tie the knt Jim and Sara had been waiting fr a lng time. They finally decided they wuld tie the knt. A B C D 84

93 Chse the picture that shws the meaning f "it's raining cats and dgs" It's raining cats and dgs Jim decided t g utside. Then it started raining cats and dgs. A B C D 85

94 Chse the picture that shws the meaning f "t see eye t eye" See eye t eye Jim and Tim have been friends fr a lng time. That is because they very ften see eye t eye. A B C D 86

95 Chse the picture that shws the meaning f "t speak f the devil" Speak f the devil Tim is saying mean things abut thers. Jim warns him by saying "Well, speak f the devil". A B C D 87

96 Chse the picture that shws the meaning f "t sit n the fence" Sit n the fence Jim is having a hard time. He is sitting n the fence A B C D 88

97 Chse the picture that shws the meaning f "t sleep with the fishes" Sleep with the fishes * Smene decided t take care f Tim. Nw he is sleeping with the fishes. A B C D 89

98 Chse the picture that shws the meaning f "t lse yur head" Lse yur head Sara has had a lng, hard day. Nw she's lsing her head. A B C D 90

99 Chse the picture that shws the meaning f "t pull a lng face" Pull a lng face When Jim sees Tim he can tell smething is wrng. "Why are yu pulling that lng face?" Jim asks. A B C D 91

100 Chse the picture that shws the meaning f "mney talks" Mney talks The businessman ften gets what he wants. That is because mney talks. A B C D 92