Assessing Multilingual Children

Transcription

1 COMMUNICATION DISORDERS ACROSS LANGUAGES Series Editrs: Dr Nicle Muller and Dr Martin Ball, Linkping University, Sweden While the majrity f wrk in cmmunicatin disrders has fcused n English ; there has been a grwing trend in recent years fr the publicatin f infrmatin n languages ther than English.'Hwever, much f this is scattered thrugh a large number f jurnals in the field f speech pathlgy/ cmmunicatin disrders, and therefre, nt always readily available t the practitiner, researcher and student. It is the aim f this series t bring tgether int bk frm surveys f existing studies n specific languages, tgether with new materials fr the language(s) in questin. We als have launched a series f cmpanin vlumes dedicated t issues related t the crss-linguistic study f cmmunicatin disrders. The series des nt include English (as s much wrk is readily available), but cvers a wide number f ther languages (usually separately, thugh smetimes tw r mre similar languages may be gruped tgether where warranted by the amunt f published wrk currently available). We have been able t publish vlumes n Finnish, Spanish, Chinese and Turkish, and bks n multilingual aspects f stuttering, aphasia, and speech disrders, with several thers in preparatin. Full details f all the bks in this series and f all ur ther publicatins can be fund n r by writing t Multilingual Matters, St Nichlas Huse, High Street, Bristl BS1 2AW, UK. COMMUNICATION DISORDERS ACROSS LANGUAGES: 13 Assessing Multilingual Children Disentangling Bilingualism frm Language Impairment Edited by Sharn Armn-Ltem, Jan de Jng and Natalia Meir t i MULTILINGUAL MATTERS Bristl Buffal Trnt

2 Cntents * c Library f Cngress Catalging in Publicatin Data A catalg recrd fr this bk is available frm the Library f Cngress. Assessing Multilingual Children: Disentangling Bilingualism frm Language Impairment/Edited by Sharn Ai^mn-Ltem, Jan de Jng and Natalia Meir. Cmmunicatin Disrders Acrss Languages Includes bibligraphical references and index. I. Armn-Ltem, Sharn, editr. II. Jng, Jan de (Assistant prfessr f language and cmmunicatin), editr. III. Meir, Natalia, editr. IV. Series: Cmmunicatin disrders acrss languages. [DNLM: 1. Language Develpment Disrders diagnsis. 2. Child. 3. Language Develpment. 4. Multilingualism. WL 340.2] RJ496.L35 ' '855-dc British Library Catalguing in Publicatin Data A,catalgue entry fr<this bk is available frm the British Library. ISBN-13: (hbk) ISBN-13: (pbk) Multilingual Matters ^ UIC: St Nichlas Huse, High Street, Bristl BS1 2AW, UK. tisa: UTP, 2250 Military Rad, Tnawanda, NY 14150, USA. Canada: UTP, 5201 Dufferin Street, Nrth Yrk, Ontari M3H 5T8, Canada. Website: Twitter: Multi_Ling_Mat Facebk: Blg: Cpyright 2015 Sharn Armn-Ltem, Jan de Jng, Natalia Meir and the authrs f individual chapters. s All rights reserved. N part f this wrk may be reprduced in any frm r by any means withut permissin in writing frm the publisher. The plicy f Multilingual Matters/Channel View Publicatins is t use papers that are natural, renewable and recyclable prducts, made frm wd"grwn in sustainable frests. In the manufacturing prcess f ur bks, and t further supprt ur plicy, preference is given t printers that have FSC and PEFC Chain f Custdy certificatin. The FSC and/r PEFC lgs will appear n thse bks where full certificatin has been granted t the printer cncerned. A. i c Cntributrs - C Intrductin Sharn Armn-Ltem and Jan de Jng Part 1: Syntax and Its Interfaces 1 Elicitatin Task fr Subject-Verb Agreement Jan de Jng c 2 Cntrastive Elfcitatin Task fr Testing Case Marking Esther Ruigendijk 3 Elicited^Prductirl f Object Clitics Philippe Prevst 4 Cmprehensin f Exhaustive Wh-Questins Petra Schulz, ^ 5 Sentence Repetitin Thedrs Marinis and Sharn Armn-Ltem r Part 2^) Phnlgical and Lexical Prcessing 6 Nn-Wrd Repetitin Shula Chiat, v Typeset by Deanta Glbal Publishing Services Limited. A Printed and bund in Great Britain by the CPI Grup (UIC Ltd), Crydn^ CRO 4YY.

3 vi Cntents 7 Using Parental Reprt t Assess Early Lexical Prductin in Children Expsed t Mre Than One Language 151 Daniela Gatt, Ciara O'Tle and Ewa Human Cntributrs 0" 8 Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual ^ Preschl Children 196 Ewa Haman, Magdalena tuniewska and Barbara Pmiechwska Part 3: Beynd Mdality 9 Assessment f Narrative Abilities in Bilingual Children 243 Natalia Gagarina, Daleen Klp>, Sari Kunnari, Kula Tantele, Taina Valimaa, Ingrida Balciuniene, IJte Bhnacker and Jel Walters 10 Executive Functins in the Assessment f Bilingual Children with Language Impairment J 277 Kristine Jensen de Lpez and Anne E. Baker Part 4: Frm Thery t Practice 11^ Clinical Use f Parental Questinnaires in Multilingual Cntexts 301 Laurice Taller s Prpsed Diagnstic Prcedures fr Use in Bilingual and Crss-Linguistic Cntexts 331. Elin Thrdardttir Sharn Armn-Ltem, assciate prfessr, Department f English Literature and Linguistics and The Gnda Multidisciplinary Brain Research Center, Bar-Ilan University, Israel. Her research interests are in language acquisitin by mnlingual and bilingual children with and withut specific language impairment (SLI). She is particularly interested in linguistic features which can disentangle bilingualism and SLI, with a fcus n syntax and its interfaces with mrphlgy and semantics, and in the impact f internal and external variables n success in child secnd language acquisitin. Anne E. Baker, emeritus prfessr, Linguistics Department f Language and Literature, University f Amsterdam, Amsterdam, The Netherlands. She is particularly interested in the relatinship between language and cgnitin, specifically in language develpment, multilingualism and disrders. Mst f her recent wrk cnsiders children wh are multilingual with ne f their languages being Dutch. V Ingrida Balciuniene is a lecturer in Lithuanian philllgy at the Department f Lithuanian Language at Vytautas Magnus University, Kaunas, Lithuania. Her research interests are in the area f first language acquisitin, with a special interest in cnversatin and discurse analysis. v n Language Index Subject Index Ute Bhnacker, prfessr f linguistics, Department f Linguistics and Phillgy, Uppsala University, Uppsala, Sweden. Her research expertise is mainly in first and secnd language acquisitin, bilingualism, grammar and discurse, with a special interest in the Germanic languages. She is particularly interested in children grwing up in Sweden with ne r several languages. Her recent wrk has expanded twards impaired ppulatins and examines typical and atypical multilingual child language develpment in a Eurpean cntext. Shula Chiat, prfessr f child language, Language & Cmmunicatin Science, City University Lndn, UK. Her interests lie in typical and atypical language develpment. Her research fcuses n the psychlinguistic

4 8 Designing Crss-Linguistic Lexical Tasks (CLTs) fr BiLingual Preschl Children Ewa Haman, MagdaLena tuniewska and Barbara Pmiechwska Overview This chapter addresses the need fr cmparable, measures f lexical knwledge in bth languages^f a bilingual child. Typically, tls designed t identify specific language impairment (SLI) d nt take int accunt whether a child is bilingual and hw, this might affect raw test scres, ften leading t misdiagnsis. Bth vcabulary size and prcessing speed can be cnfunding variables when diagnsticians attempt t disentangle bilingualism frm SLI at the lexical level. Lexical abilities can als be used as a baseline assessment f bilingual dminance/prficiency. Hence the need fr such tls as we describe here. Delayed and impaired lexical abilities are amng the earliest indicatrs f SLI (Lenard,, 1998). Children with SLI shw a delay in lexical develpment bth in terms f the verall number f wrds and in reaching lexical milestnes (i.e. first 50,100, 200 wrds; Lenard & Deevy, 2004). They als display relatively weak semantic categries (McGregr etal, 2002). Bilingual children ften have smaller lexicns in bth f their languages (Bialystk et al, 2010) when cmpared t mnlinguals. Hwever, the number f wrds in the tw languages^f a bilingual child added tgether may nt be different frm- thse measured by mnlingual nrms (Marchman et al, 2009). The prcessing lad in lexical tasks as measured by reactin time is claimed t be higher in bilinguals than in mnlinguals (Bialystk et al, 2008;.Chen, 1990rDijkstra, 2003; Khnert & Bates, 2002). At the same time, children with SLI experience reduced prcessing capabilities in cmparisn with typically develping (TD) children (Lahey & Edwards, 1996; Lahey et al., 2001; Mntgmery, 2002). ^ Lexical abilities are ptentially an early identificatin measure f bilingual SLI (Gatt et al, 2008), althugh they shuld nt be used as the Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 197 nly diagnstic variable fr this purpse (Gray et al, 1999; Spaulding et al, 2013). The assessment f prcessing speed and accuracy in lexical tasks may enhance the identificatin prcess (Perez et al, 2013). The crss-linguistic lexical tasks (CLTs) designed 11 within COST Actin IS0804 were cnceived t prvide a fully cmparable assessment f vcabulary and lexical prcessing in 34 different languages. We present the innvative methd f the CLTs ; cnstructin: a multilingual parallel task-cnstructin prcedure, which enables an bjective test f vcabulary and prcessing skills in any.pair f languages included in the prcess. The CLTs target cmprehensin and prductin f nuns and verbs. The respnse accuracy measured in the CLTs indicates the level f receptive and expressive^ vcabulary size. Measuring reactin time (i.e. cmprehensin and naming speed) prvides insight int the prcessing demands f passive and active knwledge acrss the tw wrd classes. Picture chice and picture naming were chsen as being tasks least invlving, ther types f linguistic r cnceptual skills. >. ' <;, Currently, the CLTs have been prepared fr 21 f the 34 languages and are available fr use by researchers. Their use in diagnstics will be warranted as sn as nrming studies addressing specific ppulatins f mn-and bilingual children have been cmpleted. > Backgrund Tw main phenmena characterise early wrd learning: a rapid vcabulary grwth and a significant imprvement in speed f lexical prcessing (fr vcabulary grwth, see Blm, 2000; Carey, 1978; Gldfield& Reznick, 1990; fr lexical prcessing, see Fernald et al, 2006; Garlck et al, 2001). Thus, lexical develpment seems t be best described by the fllwing variables: vcabulary size and lexical"prcessing speed. Fr children yunger than 3 years f age, standardised r nrmed vcabulary tests, such as the MacArthur-Bates ^Cmmunicative Develpment Inventries (MB Is), are available fr a number.languages (D.ale & Penfld, 2011). Hwever, fr bilingual children, after the age f 3 there are n tls t directly cmpare the vcabulary size, r the lexical prcessing speed, in bth f their languages. Bilingual children tested in nly ne f their languages appear t knw fewer wrds than their age-matched mnlinguals peers (Bialystk et al, 2010; Pearsn et al, 2006; Umbel et al, 1992), but a single language vcabulary assessment des nt fairly represent the full lexical cmpetence - in bilingual individuals. The lack f apprpriate methds fr assessing the perfrmance f bilingual children in bth f their languages makes the dmain f lexical develpment especially prne t inapprpriate assessment in these children. Mst imprtantly, reduced vcabulary size in ne f the 196

5 198 Part 2: Phnlgical and Lexical Prcessing languages f a bilingual individual may be cnfunded with delays in lexical develpment, such as is fund in mnlingual children with SLI (Lenard & Deevy, 2004). Lexical knwledge, as measured by nrmed vcabulary tests, has lw predictive value in mnlingual SLI diagnsis when used as the nly marker f language deficits (Gray et al, 1999; Spaulding et al, 2013). Hwever, it is still a factr accmpanying ther SLI markers and shuld be incrprated in the prcess f full diagnsis (Hewitt et al., 2005; Mainela- Arnld et al, 2010)... The gal f the current chapter is t present the ratinale and t give an verview f the multi-language parallel task-cnstructin prcedure. This prcedure results in the building f CLTs fr individual languages. Althugh the CLTs are nt identical in all languages (in terms f target wrds used), the cnstructin prcedure ensures that they are fully cmparable acrss all languages and within any language pair. This methd has s far been fully develped fr 21 individual languages (see Appendix C) and is under develpment in additinal languages. We expect that the CLTs may reliably demnstrate the differences in lexical knwledge (vcabulary size and lexical prcessing) between (bilingual and mnlingual ppulatins, and fr bilingual children in bth f their languages. We als prpse that the CLTs may allw us t establish the extent t which the ptential gaps in lexical knwledge between bilinguals and mnlinguals can be interpreted as situated within the range f typical develpment, and when they may indicate the risk fr SLI r ther develpmental delays in a bilingual child. The strength f this particular assessment methd is the cmparability f results between the child ; s vtw languages, assuming the languages are included in ur design (fr a full list f languages, see Table 8.1; fr the current list f CLTs versins, see Appendix C). We cntrast the way in which ur multi-language parallel task tls were cnstructed with the mre usual prcedure f adapting tls frm ne language fr use in anther (as in the case f mst Peabdy Picture Vcabulary Test [PPVT] adaptatins). Althugh the translatin f a measure initially designed fr nly ne language may seem an bvius slutin when preparing a similar tl fr anther language, imprtant arguments against this apprach are presented belw ^ The fllwing sectins describe the cnstructin prcedure f the multi-language ^parallel CLTs as develped and elabrated within the COST Actin. T the best f ur knwledge, this is the first attempt ever t crdinate the parallel cnstructin f vcabulary-assessment tasks fr such a wide range f languages. The aim was t design unifrm tls fr use in bilingual ppulatins f any language pair, frm all the languages included. The CLTs are available in tw versins: the traditinal 'paper and pencil 7 versin (which measures respnse accuracy and classifies errr types) and a cmputerised versin (which additinally measures reactin time). These three measures allw fr a deeper insight int the nature f Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 199 a

6 200 Part 2: Phnlgical and Lexical Prcessing Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 201 lexical-semantic prblems in bilingual children in general^ and in bilingual SLI in particular (Perez et al, 2013). We start by explaining ur mtivatin fr cncentrating n bilingual vcabulary assessment at preschl age. We then review measures f vcabulary currently;in use with preschlers and we review the literature n bilingual lexical prcessing during childhd. Next, we address the main assumptins behind the CLTs' cnstructin. Subsequent sectins describe the phases f the CLTs' develpment and the methd f their cnstructin. We cnclude that the multi-language parallel taskcnstructin prcedure ensures the cmparability f the GLTs results acrss varius pairs f languages, which is crucial fr the adequate assessment f lexical knwledge in bilingual children. Further studies, including nrming studies fr individual languages r language pairs, are needed t prve the diagnstic validity f the CLTs. Target age V > - In cntrast t the chapter by Gatt et al (this vlume), which cncentrates n the beginnings f lexical develpment, the fcus f the present chapter is n preschl children and mre specifically n the age f 5 years, which in mst Eurpean cuntries is just befre schl entrance age (EURYDICE at NFER, 2010; Huebler, 2010). Children's lexical knwledge at this age might directly influence their schl perfrmance. Mst educatinal institutins in Eurpe are mnlingual and use the majrity language as the language f instructin. If a bilingual child's vcabulary in the language f schling is insufficient, this can ptentially hinder, her/his cmmunicatin skills, emerging literacy and cmprehensin f written texts at schl. Thus, befre children start frmal schling, timely and accurate recgnitin, f ptential disadvantages In the dmain f lexical knwledge can, lead t early interventin, even thugh there is n simple way t clse the gap between children with substantially different vcabulary levels (Becker, 2011; Hart & Risley, 2003; Hff, 2009). ; /, An accurate diagnsis at this age has a duble advantage. First, it can lead t early interventin (i.e. maximising the child's chances-fr schl success). Secnd, it minimises misdiagnsis. Misdiagnsis may be dar&aging in tw ways. A child whse language impairment ges unrecgnised lses the benefits f early interventin. A child wh is errneusly labelled as impaired may experience impediments t her/his develpment (Paradis, 2007). Testing a child in the majrity language can nl^ assist in predicting her/his educatinal success in that language, but cannt be decisive fr assessing the risk f language impairment. Language impairment in bilingual ppulatins can be diagnsed nly if full language cmpetence, i.e. cmpetence in bth languages, is taken int accunt J (Armn-Ltem, 2011; Khnert, 2010; Paradis, 2007). J. - Measures fr assessing vcabulary size in preschlers Fr children under the age f 3 years, lexical knwledge may be assessed by questinnaire methds (e.g., parental reprts, see Gatt et al, this vlume). The availability f tests adapted frm the MB I-I and MB I-II (Fensn et al, 1993a, 1993b) fr 61 languages (Dale & Penfld, 2011) enables early assessment, even if the clinician des nt speak bth f the child's languages, because vcabulary checklists are filled in by the caregivers cmmunicating with the child in their respective languages. Althugh the MB I-III fr lder children is currently being develped in sme languages (American and British English, Danish, Dutch, Sasak, Swedish; Dale & Penfld, 2011), this is still a much less cmmn tl than the MB I-I and MB I-II and it is nt available when the child's age exceeds 42 mnths (depending n the language). Our aim is trbridge this gap by designing a set f tasks, full^ cmparable acrss different languages, fr children aged between 3 and 5 years. Bilingual children in preschl and f schl age are ften assessed n > their vcabulary size in nly ne f their languages and their perfrmance is cmpared t that f their mnlingual peers. This apprach is prblematic fr tw reasns. First, mnlinguals and bilinguals receive qualitatively different language input. Secnd, tests designed in a single language fr mnlingual speakers may in fact invlve sme interference frm ther languages (e.g. including cgnates r hmnyms). This in turn can affect the bilingual child's perfrmance n the test (Gathercle et al, 2008). Therefre, it is essential t cnstruct tasks suitable t assess the child's vcabulary size in bth languages in a cmparable manner. The tasks presented in the current chapter are an attempt t slve this issue in a systematic way fr a wide variety f languages-fey cntrlling the difficulty f test items acrss different language versins f ur test materials. S far, the research n lexical assessment in bilingual preschl children has mainly invlved specific language pairs, especially Spanish and English, and a few ther language pairs (fr Spanish-English, see Allman, 2005; Barnett &Lamy, 2006; Duursmaetal, 2007; Fernandezes/., 1992; Grman, 2012; Hammer et al, 2008; Khnert et al, 1999; Mancilla-Martinez & Lesaux, 2011; Pena et al, 2013; fr Dutch-Arabic r Dutch-Turkish, see Messer, 2010; Scheele, 2010; Van Tuijl et al., 2001; fr English-French, see Chiang & Rvachew, 2007; Elin Thrdardttir, 2011; fr English-Greek, see Lizu & Stuart, 2003; fr English-Hmng, see Kan Khnert, 2005; fr English-Mandarin, see Dixn, 2011). The aim f the present prject was t vercme the cnstraint f single.pairs f languages fr bilingual assessment. Given the variety f language cmbinatins in bilingual and ( multilingual ppulatins in Eurpe, we aimed't cnstruct quasi-universal lexical tasks which culd be freely paired frm within an extensive list f languages.

7 203 Part 2: Phnlgical and Lexical Prcessing j Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children205- Lexical prcessing in preschl age Lexical prcessing cnstitutes anther cnfunding variable in disentangling bilingualism and SLI. When cmpared t mnlingual speakers, bth TD bilingual children and children with SLI exhibit reduced lexical prcessing capabilities. Their lexical prcessing seems t be slwed dwn and. lexical access appears t be hindered (fr bilingual data, see Bialystk et al, 2008; Chen ; 1990;Dijkstra, 2003; Khnert & Bates, 2002; fr SLI data, see.lahey & Edwards, 1996; Lahey et al., 2001; Mntgmery, 2002). Hwever, bilingual and SLI individuals differ in the prfile f their lexical prcessing limitatins. SLI children display a suppressin f verb prcessing which cntrasts with an almst intact nun prcessing (Andreu et al., 2012). Bilinguals are slwer t prcess bth nuns and verbs. The specific prfile f prcessing limitatins in bilingual SLI children might be a factr ptentially differentiating bilingualism and SLI. Therefre, measuring reactin time in additin t traditinal accuracy scring in varius language tasks may significantly add t the understanding f the cmplexity and the nature f the prblems related t lexical knwledge, in particular in a bilingual ppulatin (see als Perez et al, 2013). Fr this reasn, the CLTs are prepared nt nly in traditinal paper versins, but als in cmputerised versins, which allw fr measuring the prcessing speed (reactin time) as well as prviding accuracy scres. v Basic Assumptins Underlying the Cnstructin f the CLTs Wrd classes: Nuns and verbs One f the fundamental issues f the CLTs' cnstructin was the chice f wrd categries t be included in the tasks as targets. T make the CLTs as universal as pssible, we used the tw mst cmmn wrd categries: nuns and verbs. These tw wrd classes exist in all languages (Vgel & Cmrie, 2000; Wierzbicka, 1988) and emerge early in develpment, althugh they may nt be acquired at the same time and there may be sme.crsslinguistic differences in the relative timing f their acquisitin (Gpnik et al, 1996; Tmasell et al, 1997). The lng-standirig discussin n the develpmental differences between verbs and nuns (e.g. Gentner, 1982; Gldfield, 2000; McDnugh et al,2011; Tmasell et al, 1997), including varius crss-linguistic cmparisns (Gpnik e }t al., 1996; Kaiischke et al, 2007; Tardif et al., 1997) clearly underlines that wrd knwledge assessment shuld nt be limited t nuns. SLI children seem t learn the meanings f verbs with greater effrt than the meanings f nuns. Therefre, the differences in the knwledge f nuns vs verbs are greater in SLI children than in the TD ppulatin (Andreu et al, 2012; Black & Chiat, 2003; Windfuhre* al, 2002). Hwever, Skipp et al. (2002) cmpared early wrd learning in TD and SLI children and suggested that at the age f 3, mre differences between the tw grups can be fund in nuns than in verbs with respect t grammatical knwledge. In any case, bth wrd categries differentiate TD and SLI children, which might als be reflected in bilingual ppulatins. We thus included bqth nuns and verbs when designing the CLTs. Type f knwledge assessed: Cmprehensin and prductin The assessment f tw language mdes, cmprehensin and prductin, in bilingual and language-impaired ppulatins is imprtant fe several reasns. First, cmprehensin is viewed as an accurate and representative measure f a child's lexical knwledge because it minimises the impact f ptentially interfering variables, such as lexical access and prnunciatin, and even temperamental prblems (e.g. shyness), n the child's- results (Clark, 2009). Secnd, prductin typically reveals lwer results than cmprehensin with respect t vcabulary size (Benedict, 1979; Gldfield, 2000; Harris et al, 1995; Reznick & Gldfield, 1992) and in particular is ften impaired in SLI (Capne & McGregr, 2005; McGregr et al, 2002; Messer & Dckrell, 2006). Testing bth cmprehensin and prductin gives ne the pprtunity t assess receptive, and expressive knwledge and t accunt fr the pssible difference between the tw. Althugh lexical measures cannt be used as unique predictrs in SLI diagnsis (Gray et al, 1999; Spaulding et al, 2013), lw perfrmance n these measures is cmmn in SLI (Spaulding et al, 2012). Type f tasks: Picture chice and picture naming - -, ^ The mst cmmn ways f assessing wrd knwledge in children are tasks invlving picture identificatin fr cmprehensin and picture naming fr prductin (DAmic et al, 2001; Gathercle et al., 2008; Jared et al., 2012; ICambanars et al., 2010; Mastersn & Druks, 1998; Mastersn et al., 2008). Especially useful fr children whq cannt read, these tasks are used in.mst vcabulary tests (fr cmprehensin assessment, see PPVT: Dunn & Dunn, 1997; fr prductin assessment, see Expressive Vcabulary Test [EVT]: Williams, 2006; Expressive One-Wrd Picture Vcabulary Test ^ [EOWPVT]: Gardner, 1979) and in experimental tasks (Kan & IChnert,^ 2005; Rauschke et al., 2007; Khnert et al.; 1998,1999; Messer, 2010; Scheele, 2010). Anther advantage f using picture tasks is that they minimise the - interference f ther ptentially cnfunding variables in lexical knwledge < assessment such as reasning, verbal fluency and syntactic abilities. Picture identificatin prvides the child with a limited number f pssible answers (usually fur), which are single referents fr wrds. Picture naming

8 204 Part 2: Phnlgical and Lexical Prcessing j Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children typically invlves naming a single item in the absence f any backgrund r interfering material. In this way basic knwledge f single wrd meanings can be assessed. The limitatin f picture identificatin and naming tasks is that they d nt shw whether the.child can understand r use the wrd in varius linguistic r scial cntexts. Children with SLI are claimed t have nt nly limited vcabulary.size (lexical breadth) but als pr lexical-semantic rganisatin (lexical depth); the tw factrs being strngly related (Sheng & McGregr, 2010). The lexical-semantic rganisatin may be apprached t sme extent in picture naming tasks when bth accuracy and errr types are scred. We cnsider picture identificatin and picture naming t be the best way t gauge the child's access t meanings f single wrds. This is why we used these tasks t assess lexical cmprehensin and prductin in designing the GLTs. Narrwing the list f ptential target wrds The utility f new language adaptatins f tasks initially designed fr ne language nly (Bates et al., 2003; Kauschke et al, 2007) can be questined n several grunds. One f the basic prblems is the translatin f test items. Target wrd characteristics might change in unpredictable ways frm language t language. Such changes include wrd frm cmplexity wrd frequency typical cntexts f ccurrence r the age f acquisitin (AA). These factrs may ptentially affect the, results and make them difficult t be cmpared crss-linguistically. Kauschke et al (2007) designed cmprehensin and prductin lexical tasks fr nuns and verbs in German, where the wrd length, mrphlgical cmplexity, frequency, AA 1 and picture naming agreement 2 were cntrlled. -Fr the cmprehensin task invlving picture identificatin, semantic distractrs were selected. Semantic distance between each distractr and the target wrd was cntrlled in a. preparatry study with adult native speakers assessing the distance n. a Likert-like scale. Afterwards, the authrs adapted the tasks int Krean and Turkish by translating the target wrds frm German. In the adapted -naming tasks, the target wrds were cntrlled nly fr naming agreement. Other variables (i.e.-mrphlgical cmplexity, frequency, AA) characterising the target wrds in the German versin were ignred in the adapted tasks. The adaptatins als resulted in narrwing the list f the equivalent targets in the language f adaptatin, frm 72 in the German riginal t 54 fr Krean and t 68 fr Turkish. The cmprehensin task was adapted nly fr Krean. The semantic distance between the distractrs and the target wrds was cntrlled after the translatin by adult ratings. This resulted in narrwing dwn the Krean versin by 5 items in cmparisn t the riginal German versin f 72 items. Unfrtunately, narrwing the number f items and using target wrds with different characteristics affected the-general task's,- characteristics. The lack f cntrl fr target wrds' features in new adaptatins hindered the cmparability f results in the three languages. Our design f the CLTs ; multi-language parallel task-cnstructin prcedure avids the. flaws f such traditinal translatin-adaptatins by prmting a different apprach as described belw. ' " r The CLTs' design assumes that target wrds are selected in each language accrding t the same criteria instead f being translated frm a wrd list created initially fr a single language. Therefre, the target wrds themselves may differ acrss languages but their'characteristics are stable. Fr practical reasns (i.e. preparing a cmmn set f pictures t be ptentially used in the tasks in-all languages invlved), the GLTs are based n a limited set f ptential target wrds shared acrss all languages. The target wrds fr each particular language can be selected frm this cmmn pl accrding t the rules established fr all languages in the multi-language parallel task-cnstructin prcedure. Develping a list f wrds that are equivalent in meaning acrss a wide range f languages was ur gal in the CLTs' cnstructin (Phase 1) described belw. These wrds cnstitute a surce f pssible targets fr the CLTs in each language versin and we call them the CLT-candidate wrds. T ensure the cmparability f the targets acrss all the languages, we cntrlled fr their frmal characteristics (Phase 2) and their AA in each f the languages invlved (Phase 3). The prcess f selecting the target wrds fr individual languages is described in Phase 4. The phases f the CLTs design are meant t ensure the cmparability f the results f single language assessments acrss^a wide range f languages and within any given pair f languages in the case f bilingual assessments. Pha^e 1, invlving the picture naming and rating study, was shared amng all languages and data fr each language cntributed t the verall result. The jint result f Phase 1 was the list f ptential targets, i.e. CLT-candidate wrds. Subsequent phases were intended t apply the same prcedures fr each language, but were carried ut fr each language independently. This chapter shws the ptential f the CLTs' cnstructin fp-34 languages, althugh currently the tasks are available nly fr a subset f 21 languages (see Appendix C fr the list f languages and authrs).-, CLTs' Cnstructin Phase 1: Are There Wrds Shared Acrss 34 Languages? ' The main gal f Phase 1 was t select GLT-candidate wrds shared acrss all the languages under scrutiny T this end, we designed a picture naming and rating study fr adult native speakers. We reasned that if there are bjects and actins that are easily and unequivcally named acrss the languages, their labels culd be subsequently used fr the CLTs 7

9 207 Part 2: Phnlgical and Lexical Prcessing j Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children205- cnstructin. Additinally, based n the picture rating data, we determined which pictrial style was mst universal crss-culturally and shuld be used fr the design f the new set f CLTs ; pictures. Methd r c: C 'I' Stimuli, We created a picture database with different types f pictures fr bjects and actins. The database cntained,nly pictures that had previusly been used in psychlinguistic studies in ne r mre f 15 different languages. The stimuli were gathered frm COST Actin IS0804 members 7 resurces (Russian-German, Plish, Greek, Finnish and Lebanese sets) and frm pen access surces (BOSS, SVLO and IPNP sets, see Table 8.2). Table 8.2 prvides detailed infrmatin abut the surces, the number f pictures used and the characteristics f the pictures within each surce. The database included 1024 pictures with a balanced prprtin f bjects (507 pictures) and actins (517 pictures). Each surce cntributed a number f pictures, frm 81 (7.9% f the database) up t 275 (26.8% f the database). Three surces (BOSS, IPNP and the Plish set) riginally included mre pictures than culd be cnsidered in the study. Based n the data frm previus studies which used these stimuli (i.e. the naming agreement ratings frm adults in a single language r a small list f languages), we nly included pictures with the highest within-language naming agreement scres, i.e. pictures that were named with the same wrd by mst f the participants in a naming study Even thugh these pictures had already passed the validatin prcedures fr experimental use, they had nt previusly been used in such a wide range f languages and cultures. Sme pictures frm different surces depicted the same bjects r actins, but all were included in the database. The subsets f pictures differed significantly in style (phts, black and white line drawings, different types f clur drawing). Different pictrial representatins f a cncept culd differ in their clarity and general style, which was t be assessed in the study independently frm the naming task. This enabled us t establish the mst apprpriate style f pictures. All in all, the picture naming and rating study enabled an empirical evaluatin f the relative cultural validity f different styles f picture. Participants Native-speaker judges (n=93, 81 females, 12 males) representing 34 languages were recruited frm the COST Actin IS0804 members.and their cllabratrs. All the participants were fluent speakers f English. Twentyeight (34%) identified themselves as bilingual f multilingual with English clse t native, while the remainder spke English as a freign language. The vast majrity f the judges had extensive experience^ in fields directly O IS) CNJ CO JQ. -Si "qj s.u a: 42 +JQJ t t i s cu v =3 c C3 CVI (JD ~A I r T3 E c =3 U r U sz - -a sz r Tl ZJ +J C O t a> ZJ +J O IS) +-> _ O " 'B. IS) r r E S_ V) ZJ LD Cd C3 g I c: C3 ^ =3 a- v_ a> Q- ~a -a 03 CU sz ^ r ^ -Q. -a > r ~a _aj ar "aj ^ = I i E c. r <0 ^ it: ^ a> CL) cn ~a O r E u «r r r, -a > r -a II 4-J in <P "a? 3 m "a r r i cd 5 I :zj r e " r

10 208 Part 2: Phnlgical and Lexical Prcessing j Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children O O E a g r c Z O =5, =5 a -a c 1 r c CO O.S= cn -a c R 'i u CU CL Xt- C\J C3 QJ -M LN N CO _, CVJ O cu 5 t 43 related t child language research r practice, including linguistics («=47; 57%), speech and language therapy (n=22; 27%) and -psychlgy (n=8; 10%), while nly 7% identified themselves as frm a different prfessinal backgrund (n=6). Thirty-ne f the judges (40%) "declared previus direct experience in designing picture tasks fr children. Even if nly ne infrmant was available fr sme languages, the data prvided by him/ her culd be used in the final analysis. In all but five languages (Basque, Cratian, Lithuanian, Rmanian and Spanish), we had tw t fur judges cntributing t the rating. Fr the final analysis presented here, nly the data frm the 76 judges wh rated the full set f pictures (100%) were used' (see Table 8.1 fr exact number f judges fr each language). Data frm judges wh rated at least 25% f the items were used in partial analyses nt presented here (seven judges rated mre than 25% f the items). Data frm 10 additinal judges were excluded frm the analyses due t an insufficient number f items rated (less than 13%cf items). "O a> =3 3.<u t. ^ c. C D 3 I <3 C =5 L-> v ' CNJ JD xs r 1 -Z U r <-> -Q _ - cu CN O cn O G t ; 4 1 T3 </l r u T3 t r S <D > (_) x v i c c. sz t -TO t cn C ' c OV LU > r CU n CO cn xfc\j ^ lt> LO Prcedure i An nline picture naming and rating prcedure was made accessible t the participants thrugh a passwrd-prtected website. All instructins and questins were presented in English, which was ur lingua franca fr establishing the dminant respnses in the naming task acrss all languages. The participants fllwed the prcedure (i.e. they named the pictures) in their native languages and prvided the English translatin f each wrd used. ^ The pictures were gruped int the tw-weird categries, s that infrmants assessed pictures fr nuns and verbs separately. The rder f the tw categries was randmised acrss participants. The rder f the pictures within each categry was randmised fr each participant t avid rder effects. The pictures were presented ne by ne and each was accmpanied by fur questins, tw relating t its'label and tw relating t its style. The image was placed in the tp left crner f the screen (ccupying abut a quarter f the screen, uspace). The^questins were displayed n the righthand side f the screen. Picture naming. First, the participants judged whether the picture easily evked a wrd in their native language. They rated the pictures n a 5-pint scale ('n, nt at all'; 'n, but fhave sme vague ideas'; 'yes, it evkes several wrds different in meaning'; 'yes, it evkes several wrds similar in meaning'; 'yes, it evkes ne wrd'). Then, they named the picture with the first wrd they culd think f. They prvided a wrd in their native language first and then translated it int English. The participants were encuraged t use a dictinary if they were nt sure abut the best English equivalent. Picture-style rating. The next tw questins were aimed at assessing the style f the picture. The participants were asked whether the picture was K

11 210 Part 2: Phnlgical and Lexical Prcessing an accurate example f the bject/actin they had just named and whether the general style f the picture (manner f drawing, shapes, clurs) wuld be suitable fr children in their cuntry Results btained fr these tw questins were then used t determine the general style f the new pictures t be designed fr the CLTs and t give specific instructins t the illustratr wh was t draw them (see /Designing and Sharing Pictures fr CLTs 7.belw).-Fr bth questins referring t the style, a 4-pint scale (Very gd'satisfactry''a bit strange'very strange 7 ) was used. Results We cnducted tw separate analyses. First, we determined which wrds evked unifrm answers acrss mst languages. Secnd, we selected the picture style that was the mst suitable acrss cultures. Naming The aim f the first analysis was t identify the wrds which wuld be mst suitable fr the CLTs cnstructin. 3 V\fe wanted t avid bjects/ actins which did nt evke a wrd in a given language r evked mre than ne wrd f different meanings. We expected that pictures named cnsistently with just ne wrd r with several wrds similar in meaning were ging t be mst suitable fr the purpses f vcabulary assessment. First, based n the English-language equivalents prvided by the judges, we identified a dminant naming respnse fr each f the 1024 pictures, that is, the name that was prvided by the largest number f participants. Then, fr each picture, we calculated the fllwing tw indices: a Dminant Name Index (DNI) and a Meaning Availability Index (MAI). DNI. The DNI fr each picture was the prprtin derived frm the number f times the dminant English equivalent was iised fr the given picture, divided by the ttal number f respnses. Fr instance, a picture presenting a pharmacy was named a pharmacy by 53 ut f 76 judges. 4 Thus, the DNI f this picture was 0.70 (53/76). MAI. The MAI fr each picture was the prprtin derived frm the number f ratings 'evking ne wrd 7 r 'several wrds similar in meaning 7, divided by the ttal number f ratings. Fr example, 50 judges assessed the same picture f a pharmacy as 'evking ne wrd 7 and 13 ther judges decided that it 'evked several'wrds similar in meaning 7. Thus, the MAI f this picture was 0.83 (63/76). Bth the DNI and MAI values fr all CLTcandidate wrds are given in Appendix A. The DNI and MAI indices were in general higher fr pictures featuring bjects (evking nuns, DNI: M=0.77, SD=0.22, MAI: 0.92, SD 0.12) than fr pictures featuring actins (evking verbs, DNI: M=0.64, SD=0.22, MAI: M=0.77, SD=0.16). These results are in line with the well-established claim that nuns are much mre stable N crss-linguistically than verbs (Gentner, 1981, 2006). This shws that in Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 211 a wide crss-linguistic cmparisn: (1) the bject pictures,are named in a mre unifrm way than the actin pictures.and (2) the bject pictures seem t evke fewer wrds f different meanings than the actin nes. Thus, when selecting wrds that evked the mst unifrm respnses in the picture naming task, we adpted different threshld criteria fr nuns and verbs t reflect the specificity f these tw wrd categries. We selected 158 nuns and 142 verbs as,clt-candidate wrds, frm which target wrds fr varius language versins f CLTs culd be chsen in Phase 4 (see Sectins 'Nuns 7 and 'Verbs 7 belw). Nuns. The CLT-candidate, nun list was cmprised f the naming respnses fr pictures with MAI scres f at least 0.99 (see Appendix A). Out f 507 pictures, 232 representing bjects met this criterin. Hwever, the final nun sample cnsisted f 158 wrds', because 49 bjects were represented at least twice in the set f pictures selected. We als cntrlled fr the DNI f the selected pictures. Out f the ttal f 158 items, 34 (22%) resulted in DNI=1.00 and 112 pictures (71%) had a DNI higher than Only six items had a DNI lwer than The DNI fr the six pictures was s lw because they were named with tw cmpeting wrds (clse synnyms) f similar frequency (e.g. a xpicftire illustrating a cap was labelled as cap by 38% f judges and as hat by 33% f them). Verbs. The CLT-candidate verb list was cmprised f all the verbs which met a minimum criterin f MAI >0.90. Out f 517 pictures, 269 met this criterin. These 269 pictures crrespnded t 142 wrds, since there were sme illustratins depicting the same activity and 66 f the chsen verbs were represented by at least tw pictures. The majrity (56%) f the selected verb pictures had a DNI higher than Only 17 images (12%) resulted in a DNI lwer than 0.50 (see Appendix A). Once again, this situatin was assciated with the ccurrence f mre than ne dminant synnymic answer fr sme pictures (e.g. the picture illustratin fr calling was named t talk by 21% f the judges, and t call, t phne x t talk n the phne by 19%, 19% and 10%, respectively). The CLT-candidate nun and verb lists were used in Phases 2-4 f the CLTs design. In Phase 2, their phnetic and mrphlgical characteristics were btained fr each language under scrutiny In Phase 3, we assessed the AA f the wrds fr each f the languages. Finally, in Phase 4, we selected target wrds fr each language. Picture-style rating Belw we reprt the'results btained thrugh tw picture-style questins. Participants 7 answers were analysed acrss the different surce - picture sets and wrd categries. Since nt all picture sets cntained bth (bjects and actins, we reprt the results fr the tw categries separately. Mean respnses fr individual pictures between 1 and 2 (crrespnding t respnses: Very gd 7 and 'satisfactry 7 ) were cnsidered as meeting the

12 213 Part 2: Phnlgical, and Lexical Prcessing ^ Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 219 criteria f sufficient picture accuracy (first questin) and adequate style (secnd questin). Nt surprisingly the ratings fr picture accuracy were significantly crrelated with the ratings fr general style (bject pictures: r=0.86, p<0.0v, actin pictures: r=0.74, p<0.01). Fr general style, the mean ratings fr all sets were belw 2, except fr ne. This indicated that mst sets were rated as at least satisfactry in style (1 crrespnded t Very gd' arid 4 t Very strange 7 ). Only the black and white drawings frm the IPNP set (verbs nly) btained a mean rating f 2.19 (SD=0.29).,.. Fr the bject pictures, phts withut backgrund (the BOSS set, M=1.49; SD=0.37) and clured pictures withut shadwing (Finnish set ; M=1.49; SD=0.35) btained the best mean ratings. Next came the SVLO set (see Table 8.2) (clured pictures withut shadwing, M=1.51; SD=0.29). Other clured pictures r pht sets (the Plish, Greek and Russian-German sets) were rated frm M=l.67 t M=1.71. The lwest rating was btained fr the black and white line drawings (the Lebanese set), with M=1.97 (SD=0.27). Fr the actin^pictures, the highest mean rating was btained with the Finnish set (M=1.26; SD=0.17) and the secnd highest were the waterclur pictures (Plish set, M=1.50; SD=0.25). There were differences between.'the ratings f ther sets. The Russian-German set (clured pictures, sme shadwing) was-rated at almst the same level as the Lebanese set (M=1.95; SD=0.32), and similarly the phts with backgrund (Greek set, J M=1.92; SD=0.024),The black and white line drawings scred lwest (IPNP: M=2.19; SD=0.29; Lebanese set: M=1.94; SD=0.20). Overall, the clured pictures withut shadws were rated best fr bth bjects and actins. This became ur default drawing style fr the new picture database. The DNI f individual pictures was used t infrm the.artist abut hw accurate they were assessed t be (described in detail in 'Designing and Sharing Pictures fr CLTs' belw). CLTs' Cnstructin Phase 2: CLT-Candidate Wrds - Assessing Their Characteristics Acrss 34 Languages - Our aim in^phase 2 was t determine hw mrphlgically and phnetically cmplex the CLT-candidate wrds were. Belw, we summarise the methd established by ur team t assess wrd cmplexity. First, n the basis f literature reviews and discussins within the COST Actin, we established a list f wrd features that can affect the accuracy and reactin time in picture naming and wrd cmprehensin tasks. These wrd features were then used as input fr calculating the Cmplexity Index (CI) fr each CLT-candidate wrd in each language under scrutiny. c ^ Expert linguists (ne persn per language) prvided infrmatin n characteristics f the CLT-candidate wrds by filling in specially designed questinnaires. We inquired abut the phnlgical and mrphlgical structure f the chsen wrds. In the phnlgical dmain, we asked abut the wrd length in-phnemes (Mrrisn et al, 1992), the presence f cnsnant clusters (Brwn & Watsn, 1987; Santiag et al, 2000) and the initial fricatin (Barca ^ al, 2002; Brwn & Watsn, 1987). In the mrphlgical ^dmain, we cllected infrmatin abut wrd frmatin in rder t identify items frmed by derivatin and thse frmed by cmpunding (Baayen et al, 2006; Juhasz et al, 2003; Zwitserld et al, 2000). Additinally, we gathered infrmatin n wrd etymlgy t establish whether a particular wrd was a recent lanwrd. Experts als prvided infrmatin n children's direct expsure t wrd referents (e.g. snwmen are nn-existent in Israel, hence a child's direct expsure t these entities may be very limited there). We then used this infrmatin t cmpute the CI fr each wrd in each f the languages. The CI was calculated in the same way fr nuns and verbs acrss all the languages invlved in rder t enable cmparisns between individual wrds, as well as wrd classes within and acrss languages. The CI was derived frm the fllwing frmula: 1 CI=L+SLP+B+D+S+P+E+F+I+InitC+InterC where: L means being a recent lanwrd (1 pint if the wrd is a lanwrd, 0 if it is nt) SLP means dubled standardised length in phnemes in a given language, calculated separately fr nuns and verbs 5 B means number f rts (cnsequently, mre than 1 pint fr the cmpund wrds) D means being a derived wrd (1 pint if the wrd is derived, 0 if it is nt) S, P means suffixes and prefixes (1 pint fr each suffix and prefix) E means expsure t the referent (0 pint if the bject/actin is available t children based n direct experience in a given culture, 1 pint if it is nt) F means subjective frequency f expsure (0.5 r 1 pint fr rare^bjects, 0 fr cmmn nes) T means initial fricatin (1 pint if the wrd begins with fricatin, 0 if it des nt) ^ y Init'C means initial cnsnant clusters (I pint if the wrd begins with a cnsnant cluster) InterC means internal cnsnant clusters (1 pint if the wrd cntains a cnsnant cluster) Fr instance, the CIs f the English wrds ball and blackbard equal and 7.86, respectively. The wrd ball is relatively shrt (shrter than mst ther nuns; SLP=-1.59; as SLP is a standardised measure,

13 214 Part 2: Phnlgical and Lexical Prcessing its values are negative fr all wrds shrter than average and psitive fr all wrds lnger than,average) and frmed with ne rt nly (s B=l). It is neither/a lanwrd nr a derived wrd and it des nt cntain any initial fricatin, initial r internal cnsnant clusters, and is a familiar bject in the.children's envirnment. The CI f this wrd is the sum f SLP and B (all ther cmpnents are 0), which is = The wrd blackbard is nt a lanwrd (L=0) ; it is relatively lng (lnger than mst ther nuns; SLP=2.86), cnsists f tw rts: black and bard (B=2), is nt derived-(d = 0), des nt begin with an initial fricatin (1=0) but Cntains bth initial and internal cnsnant clusters (InitC=l, InterC=l). Blackbards are available fr children in the British culture i c (E=0) but were assessed as rare in their direct experience (F=l). Thus, the CI f the wrd blackbard is 2.86 (SLP)+2(B)+l(InitC)+l(InterC)+l (F)=7.86. Fr the design f CLTs ; the CI is used as nly ne f tw indicatrs f wrd difficulty which can ptentially affect the accuracy and latency f children's respnses. The ther factr is the AA (see next sectin describing Phase 3 f"clts ; cnstructin). We explain hw the CI value is included in the CLTs' design in the sectin in the sectin 'CLTs' Cnstructin Phase 4: Selecting Targets fr Each Language 7. CLTs' Cnstructin Phase 3: Are CLT-Candidate Wrds Acquired at a Similar Pint in ALL Languages? AA Study Anther factr which we chse as ptentially affecting children's perfrmance in lexical tasks was. the AA f wrds, i.e. the estimated age at which children acquiring a given language start t cmprehend a wrd (e.g. Stadthagen-Gnzalez & Davis, 2006). Althugh there are many studies assessing bth subjective AA (estimatin f AA n the basis f adults 7 reprts: answers t questins like Svhen did yu learn this wrd?') and bjective AA (the actual measurement f the age at which at least 75% f native speakers recgnise a given wrd in varius lexical tasks, mstly in picture naming), the data in these studies were gathered in varius 'designs'and fr varius,sets f wrds, which makes direct cmparisns difficult (Perez &: Navaln, 2005). Previus studies shwed that there is a strng relatinship between bth measures f AA (Gilhly & Gilhly 1980; Mrrisn et al., 1997; Ymd et al., 2000). Mrever, the AA data are nt available fr all languages invlved in the CLTs. Thus, COST IS0804 launched a hew study n subjective AA t btain fully cmparable data fr all the languages under scrutiny and fr all the CLT-candidate wrds chsen in Phase 1 (fr the results f this prcedure, see tuniewska et al, submitted). Belw, we briefly present the methd used^t- btain the Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 215 ~ ' L -, subjective AA data and we explain hw these data were applied in the cnstructin f the CLTs. ^ Methd: Subjective AA measurement Given that ur study was ging t be cnducted in multiple languages, we chse the subjective AA measurement as it had previusly prved t be well adapted fr multilingual cntexts. Our prcedure clsely resembles the /ne used in previus studies (e.g. Carrll & White, 1973; Gilhly & Lgie, 1980;-Mrrisn et al., 1997). We intrduced ne change: we applied a different scale frm that cmmnly used in earlier studies. The majrity f AA studies used either a 9-pint scale ranging frm 1 (age 2) t 9 (age 13+) prpsed by Carrll and White (1973), r a 7-pint scale ranging frm lf(age 0-2 years) t 7 (age 13+) intrduced by Gilhly and Lgie (1980). These ranges are widely accepted as methds f representing age in the AA literature. Hwever, the scale intrduced by us is mre precise in the age intervals relevant fr CLTs' cnstructin. It ranges frm 1 t 18, with steps representing the exact age in years. This scale was bth easier t understand by participants and mre accurate fr AA measurement f early wrds. The results btained with the new AA scale can als be easily transfrmed and s they can be cmpared with the results f any previus studies, in cntrast t data cllected with scales which include pints representing intervals lnger than ne year. Furthermre, mst f the previus AA studies relied n paper-based questinnaires (e.g. Carrll & White, 1973; Gilhly & Gilhly, 1980; Mrrisn et al., 1997). We launched an electrnic versin f the study. The nline prcedure enables the fast and efficient, recruitment f participants frm many different cuntries and speaking different languages. Fr each language, a separate versin f the AA assessment was prepared by the members f the COST Actin. Participants Participants were recruited by the members f COST IS0804. Samples fr the individual languages rangerd frm 20 t 136 adult participants (M=33T62, SD =23.00). There was a strict minimum f 20 participants per/lahguage. They were all adult native speakers f the languages selected wh vlunteered t,participate in the study. Their demgraphics (age, gender, educatin, number f children) as well as self-reprted data abut language skills and use (number f knwn and used languages) were cllected. : Prcedure Participants were asked t dwnlad a questinnaire in their mther tngue frm the prject's website ( and then fill

14 217 Part 2: Phnlgical, and Lexical Prcessing ^ Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 219 it in. The questinnaires included detailed instructins, questins abut demgraphic data and, mst imprtantly, lists f nuns and verbs fr the AA estimatin, Subjects were asked t estimate the age at which they learned the wrd (i.e. started t understand it), by typing a number between 1 (if they thught they had learned the wrd when they were 1 year ld^and 18 (if they thught they had learned the wrd when they were 18 r lder). Each participant was presented with a different randm rder f wrds t minimise the risk f latent rder effects. Nuns and verbs were presented separately. Each participant was given the full list f all 300 Clfl-candidate wrds. The task duratin was abut half an hur. This prcedure was applied t all languages included in this phase apart frm Nrwegian. The Nrwegian AA study was cnducted with the very same target questins 'but with a different nline prcedure (Lind et al., 2013, 2015; Simnsen et al., 2012). Hwever, cmparisn f the results reveals ; that, the Nrwegian, data resemble all ther data: crrelatin cefficients between the AA rates in Nrwegian and in ther languages included in the sjtudy d nt differ frm thse btained in ther pairs f languages tested. J?r three f the languages (Hebrew, Irish and Luxemburgish) the nline prcedure failed fr practical reasns (prblems with recruitment). That is why the questinnaires were printed ut and delivered in paper versins in a frm exactly matching the questinnaires accessed frm the website in these languages. Results Preliminary utcmes (Haman et al, 2011; Luniewska et al., 2012a, 2012b, 2013, 2014; Suthwd et al, 2011) suggest that there is a strng relatinship between the estimated AA f the wrds amng all the languages included in the study, and that the majrity f the GLT-candidate wrds are typically assessed t be acquired befre the age f 8 years (fr detailed results, see Luniewska et al, submitted). The wrds acquired earliest in a sample f 20 languages 6 are: nse (M=2.24, SD=0.63), ball 1 (M=2.24, SD=0.52) and bed{ba=2.2a, SD=0.56), and the wrds acquired latest are: cmputer (M=8.46, SD=1.95), t surf (ba=8.12, SD=1.91) and t hitchhike, (M=7.93, SD 1.43). The verall mean f AA in these 20 languages differs between nuns (M=3.21, SD=1.01) and verbs (M=4.86, SD=1.48). Overall, the results suggest that the CLT-candidate wrds are acquired in a similar rder and at apprximately the same time in all languages studied. The results fr individual languages were emplyed in the next phase f CLTs ; cnstructin. The AA value, i.e. the mean per item per language, was used tgether with the CI t determine the difficulty level accrding t which the target wrds culd be selected fr each versin f the CLTs. The precise applicatin f these tw factrs t select the targets is described belw. ^ CLTs'Cnstructin Phase 4: Selecting Targets fr Each Language c The CLTs are a series f fur sub-tasks fr: the cmprehensin f nuns, the' cmprehensin f verbs, the prductin f nuns and the prductin Lverbs. The cmprehensin tasks use the picture identificatin prcedure and the prductin tasks use picture naming. The number f targets fr each task and each wrd categry is limited t 30 items per categry. This number ensures that bth the cmprehensin and the prductin f nuns ana verbs can be assessed with satisfactry accuracy and that testing can be accmplished within a reasnable time/ Additinally, there are tw training items fr each task/wrd categry. Belw, we describe detailed criteria fr the selectin f targets fr prductin and cmprehensin and als f distractrs fr wrd cmprehensin. We used three distractrs fr each target. Presenting tfie child with fur pictures (ne target plus three distractrs) fr each item minimises the prbability f randm chice and at the same time \des nt vertax the perceptual lad. This fur-picture slutin is used in many standardised wrd cmprehensin tests (e.g. PPVT: Dunn & Dunn, 1997; BPVS: Dunn et al, 2009; OTSR: Haman & Frnczyk, 2012). Additinally, target wrds fr prductin are chsen frm distractrs in the cmprehensin task. The targets fr cmprehensin and prductin are thus matched as clsely as pssible in accrdance with the criteria described belw, t guarantee a similar level f difficulty between the tw tasks. r; During Phases 2 and 3 f the CLTs ; cnstructin, a CI (Phase 2) and a value f AA (Phase 3) was assigned t each CLT-candidate wrd in each language^ Accrding t these values fr each particular language, all CLT-candidate wrds were nw assigned t ne f fur difficulty levels in a 2x2 design (CI: lw/high; AA: earlier/later). Fr each task and wrd categry, the same number f targets fr each difficulty level (cmbined CI and AA) are t be selected. This will enable subsequent assessment f whether these variables affect the children's results in an ANOVA design. Table 8.3 presents the distributin f targets acrss tasks, wrd categries and difficulty,levels. ^ Additinally, t ensure' semantic variety, nuns and verbs were assigned t ne f three brad semantic categries. Nuns were divided int animate natural kinds (e.g. butterfly, frg, tiger), inanimate natural kinds (e.g. cucumber, clud, leaf) and artefacts (e.g. drum, lamp, snwman) (Keil, 1989). Verbs were divided int physical actins perfrmed by humans (e.g. t laugh, t paint, t peel), actins perfrmed by animals (e.g. t bark, t hatch, t sting) and states r unintentinal actins (e.g. t bil, t drip, t rain). Thus, fr each difficulty level, we aimed at including targets frm different semantic categries.

15 218 Part 2: Phnlgical, and Lexical Prcessing ^ Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 219 > - Z2 ', a> -a. Z5 I Q. -a ; c r c cu i CL _ E fc S -a r < u <4- O a> < -a X -a a> c Q. E c -a aj ^ ta> 5 g <"J >> CO * a> u i.2 ^ - P cu In the prcess f target wrds selectin, we used the CI and AA values tgether with the infrmatin n whether the wrd frm is a lanwrd in a given language and whether and hw f ten the children may have had direct experience with the wrd's referent (in a given language/cultural cntext). At first, 32 pairs f target wrds fr prductin and cmprehensin fr each wrd class were determined, with exactly 8 pairs in each level f difficulty within each wrd class. The wrds in each pair were taken frm the same semantic categry and matched fr their CI and AA values. Whenever pssible, lanwrds and wrds with referents nt directly accessible t children were avided as targets. Then, the additinal tw distractrs were chsen frm the set, f CLT-candidate wrds fr each f the cmprehensin targets n the basis f difficulty level and semantic dmain. We avided jphnlgical distractrs such as wrds phnetically similar t the target, < e.g. dll-dg r fly-flag in English, as well as perceptual distractrs such as wrds with pictures similar in shape t the target picture. Fr example, a/picture f a leaf similar in shape t a feather shuld be excluded as a distractr fr the target wrd feather. Phnlgical distractrs were nt cnsidered at all since fr a wide range f languages invlved in the CLTs, it wuld nt be pssible t systematically select phnlgical distractrs fr each target wrd frm a limited set f CLT-candidate wrds. Fr each language included in ur sample, the prcedure f selecting targets and distractrs is exactly the same. Detailed instructins and all basal data (CI, AA values, etc.) are available in an MS Excel file that supprts the autmatic srting f wrds, cntrls fr the number f targets/distractrs at each f the difficulty levels and mnitrs the semirandm placement f the target wrds in the cmprehensin tasks. Fr each language, the natinal teams in the COST Actin cmpsed the list f targets. These teams are the authrs f each language versin f the CLTs. We list the versins already available tgether with the names f authrs in Appendix C. Administering Crss-Linguistic Lexical Tasks: The Methd /.1 ^ : - :. In this sectin, we describe the intended target grups and the prcedure fr testing children using the CLTs. The paper and pencil versin f the CLTs used fr assessing accuracy is cnsidered t be the basic ne. This is due t expected cnstraints in testing fr diagnstic purpses, such as speech and language therapists' limited access t prfessinal electrnic equipment in many cuntries. The electrnic versin that can be used fr assessing bth, accuracy and latency is intended mainly fr research purpses, but diagnstic use is nt"excluded. It shuld be emphasised that the diagnstic use f the CLTs will nly be pssible if the tasks are nrmed in relevant languages with the relevant grups targeted. Extended research is needed

16 220 Part 2: Phnlgical and Lexical Prcessing <0 ^ \ ^ fr each language/language pair befre any use in clinical r diagnstic setting is, pssible., v Expected.target grups/participants The CLTs were designed t assess lexical knwledge befre children start frmal schling. It is expected that, fr each language, baseline data'fr mnlingual 5-year-ld children will be gathered. Our main targeted grups: TD bilingual children, mnlingual SLI children and bilingual SLI children, are ging t be cmpared t these baseline grups^ Each grup shuld cnsist f at least 20 children characterised by hmgeneus backgrund variables (sci-ecnmic status [SES], expsure t languages, etc.). Prcedure Jtems within each f the fur tasks are randmly rdered with the ^tw^ simplest practice items at the beginning f each task. The pictures fr ' the)cmprehensin tasks are als semi-randmly assigned t ne f fur psitins (upper/lwer and left/right; see Appendix B fr sample picture bards). This is dne accrding t three rules: (1) each f the fur psitins gets quarter pictures fr target wrds; (2) target pictures cannt be lcated in the same psitin in mre than three cnsecutive picture bards; (3) acrss seven subsequent items, target pictures appear in each psitin at ' least nce. The instructins as well as the answer sheets were riginally prepared in English and were adapted t each f the languages invlved (fr the list f authrs, see Appendix C). Since the cmprehensin tasks include distractrs. which are used in the prductin tasks, a ptential influence f. task rder n the results cannt be ruled ut. Thus, fr research purpses the rder in which the tasks are administered shuld be balanced acrss participants. We recmmend that the cmprehensin tasks (fr nuns and Verbs) be administered cnsecutively, and that the prductin tasks be administered ne after the ther as well.-the rder f the tasks invlving nuns and verbs is balanced acrss participants. All fur tasks can be/delivered in a single sessin r with breaks in between. N break shuld be allwed within a task. The estimated ttal time f testing fr all fur tasks is up t 20 minutes. Children shuld be tested in a mnlingual mde with the experimenter cmmunicating with them in ne language. Althugh spntaneus answers in ther languages are nt rejected, they shuld nt be encuraged. Paper and pencil versin Fr the cmprehensin tasks, each item is cmprised f a fur-picture bard (each representing ne target wrd and three distractrs). Pictures p ' Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 221 in-black frames (7.5x7.5 cm) are symmetrically lcated n an A4 page and,; numbered clckwise (hrizntal view). Fr the prductin tasks, single pictures in black frames (7.5x7.5 cm) are centrally lcated n an A5 page (hrizntal view) (see Appendix B). Each bard is accmpanied by a questin prmpt. Fr the cmprehensin tasks, the questin is 'Where is... v;(target nun)? ; (e.g. 'Where is the gate? 7 ) r 'Wh is... (target verb in relevant frm)?' (e.g. 'Wh is kissing?'). Befre starting the actual testing sessin, the child is infrmed that she/he is ging t watch a series f pictures and will be asked questins abut the pictures. The child is als asked whether she/he agrees t attend the^ sessin. Having agreed, the child is instructed t pint t the picture which ges best with the wrd in the prmpt questin (fr cmprehensin tasks), r t answer the questin by naming the picture (fr prductin tasks). Fr the latter tasks, single wrd respnses are encuraged ('ne wrd is enugh'). Fr the first tw practice items, the child is given feedback when her/his answer is nt crrect. Fr all the remaining items n feedback is given. Fr each answer, the experimenter prvides nly the minimal natural reactin t sustain interactin, with n. vert psitive r negative feedback t the child's answers ('aha', 'kay', etc.). At the end f all 32 items f the task, the next task is delivered r a break is ffered if needed. After all fur tasks, the child is cngratulated and is thanked fr her/his participatin. Fr the cmprehensin tasks, the experimenter ntes the child's respnse n the answer sheet (the number f picture pinted t); n recrding is required. Fr the prductin tasks, an audi recrding is required. The experimenter ntes the child's respnse n the answer sheet nly t the extent that it des nt affect natural cmmunicatin. The recrding is then cnsidered t be the basis fr transcriptin and scring. Ntes n the answer sheet can be used t supprt transcriptin if needed. During the testing with bth types f tasks, the experimenter is asked t make a nte f any atypical behaviur frm the child. r /. '. Cmputerised versin J f \ '. -. jthe pilt electrnic versin (develped as a desktp applicatin in C# using.net Framewrk 4.5; Etenkwski & Luniewska, 2012) was prepared s that it resembles the paper versin as clsely as pssible. The picture bards are presented n a tuchscreen in the same rder and layut as in the paper and pencil versin. The prmpts are pre-recrded and a strict timing schedule is fllwed. Prmpt questins are played befre the frthcming picture bard becmes visible (with a 100 ms lag frm the ffset). Fr cmprehensin, children are suppsed t tuch the screen while pinting t the picture f their chice. The next questin is played autmatically after a delay f 150 ms frm the child's respnse. The child's picture chices and

17 222 Part 2: Phnlgical and Lexical Prcessing ^ v. Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 223 her/his reactin times (the latency frm the mment f the picture bard emergence t the mment when the child tuches the screen) are stred autmatically. Fr prductin, children are suppsed t watch the pictures n the screen and answer the pre-recrded questins. The next picture bard is presented n the screen after the child has finished respnding. The child's respnses are audi recrded. Expected utcmes: Preliminary results The methd f CLTs' cnstructin described in this chapter was designed t ensure the effects we list belw. At the mment, these are still anticipated results f future research althugh partial data already available supprt ur predictins (Haman & Luniewska, 2013). The CLTs are expected t differentiate between grups f mnlingual, bilingual, TD and SLI children. The accuracy and reactin times fr the cmprehensin tasks are expected t exceed thse fr prductin. It is als expected that the accuracy scres n the nun tasks will exceed the scres n the verb tasks. It is predicted that there shuld be an interactin f the wrd categry and task type fr SLI children. While bth nun and verb cmprehensin tasks shuld reveal the gap similar t TD children, the difference between the tw wrd categries in prductin shuld be significantly wider as verbs are particularly difficult fr SLI children. If children ytmger than 5 years f age (baseline) are tested, the accuracy f the CLTs is expected t increase with the child's age. We als expect the level f difficulty r ther effects f cmplexity and AA t influence the results. We predict that wrds with lw CI values will be understd and prduced earlier and prcessed quicker. An analgus effect is expected fr the AA factr: the wrds with a lw AA index are expected t be understd and prduced earlier than thse with a high AA index. S far, the CLTs versins listed in Appendix C have been pilted and administered t sme grups f mnlinguals and bilinguals, TD children and thse diagnsed fr SLI (Haman & Luniewska, 2013). Hwever, results btained t date have been incmplete (sme grups were tested in ne language nly, fr sme languages nly TD children were tested, etc.). Hence, they cannt be used as cnclusive evidence fr the general findings we list abve. Presentatin f the empirical results ges beynd the aims f the current chapter. The actual results fr the relevant languages and targeted grups will be published in J a series f separate publicatins. Designing and Sharing Pictures fr CLTs The results f the picture naming and rating study (described in 'CLTs' Cnstructin Phase 1: Are There Wrds Shared Acrss 34 Languages?' abve) indicated that the mst widely accepted style f pictrial stimuli are hmgeneusly clured drawings with n shadwing. Using this bservatin, a new set f pictures has been designed exclusively fr the CLTs. The electrnic CLTs' picture database cmprises pictures fr all 300 wrds frm the CLT-candidate wrd list (158 nuns and 142 verbs). Fr sme f them, several variants are available. There are 416 distinct.jpg files f unifrm parameters in the database, including 361 base pictures and 55 variants. The artist respnsible fr the picture design (wh is experienced in designing varius types f pictures fr child language research) was infrmed in detail abut the aims f the prject. In additin t the list f CLT-candidate wrds t be illustrated, she received the relevant pictures rated and named in Phase 1 (see 'CLTs' Cnstructin Phase 1: Are There Wrds Shared Acrss 34 Languages?') tgether with the ratings fr their ' accuracy and style. She was instructed abut the preferred style f pictures (clured line drawing withut shadwing) and abut the additinal criteria listed belw The additinal criteria ensured the crss-cultural fairness f pictures. \They were adpted accrding t the cnclusins resulting frm discussins in the COST Actin meetings: (1) Balancing the number f male and female characters perfrming actins (verb pictures). (2) Aviding gender steretypes (i.e. wmen perfrming typical husehld actins and men technical and prfessinal activities) with the.reservatin that the gender f the character perfrming a particular actin shuld nt be cnspicuusly incngruus (e.g. a wman shaving her face). (3) "Aviding racial steretypes (i.e. aviding pictures in which nly Caucasian characters perfrm actins) with the reservatin that fr ' cultures where ethnic diversity is scarce the final set f pictures shuld nt include many examples f characters frm different ethnic grups. 1 These criteria were fulfilled and a balanced number f male and female characters.within the whle database was assured; 32% f pictures shwed female characters, 40% male characters and 28% were gender neutral, i.e. witlfr bth genders represented in ne picture, r n gender infrmatin prvided, such as when nly hands are visible. In rder t avid ethnic steretypes, variants f sme pictures were created t represent distinct ethnic grups (African, Chinese, Asian, Indian). The CLTs picture database includes 34 ethnic.variants f actin pictures invlving human characters (24% f all verbs) and 8 single actin pictures with figures frm nn- Caucasian ethnic grups (6% f -all target verbs). Cnsidering that 16 actin pictures (11% f all verbs) d nt invlve any human character (pictures fr verbs like t rain, t burn), it turns ut that ne third f the actin pictures

18 224 Part 2: Phnlgical and Lexical Prcessing Is invlving human characters is available with nn-caucasian peple. We argue that the CLTs ; picture database is by and large ethnically apprpriate fr use in varius Eurpean cntexts. All the. pictures were reviewed by a panel f 15 COST Actin members frm 9 different^untries. Fllwing their suggestins, 41 pictures were mdifiedx14% tall 300 CLT-candidate wrds). Sme f these mdificatins resulted in adding, new pictures t the database. Fr example, pictures fr bth red and grey squirrels are available; as well as tw differently shaped tree leaves r tw types f scales., The electrnic CLTs' picture database was funded by the Ministry f Science and Higher Educatin (Pland) with the cpyright held by the University f Warsaw (Pland). Access t the database fr research purpses as described in the present chapter is given t all interested parties n the basis f a free license agreement between the University f Warsaw and relevant institutins vvhere members are invlved in cnstructing the CLTs language versins. The CLTs language versins With full instructins and pictrial material are, ging t be available fr research purpses via the COST Actin IS0804 website fr the child language research cmmunity. Cnclusins (The CLTs were designed as a set f lexical tasks assessing cmprehensin and prductin f nuns and verbs in bilingual children and children with SLI. This assessment was required t be fully cmparable between the tw languages f a bilingual child. T meet this need, the CLTs were designed in separate languages, but they can be paired acrss a wide range f languages withut changing the CLTs ; characteristics. The different phases f the CLTs ; develpment were cnceived t guarantee the stable characteristics f tasks acrss all languages included in the sample (see Table 8.1). The results available s far fr mnlingual, bilingual, mn-sli and bi-sli children cnfirm that the. CLTs can differentiate between these grups (Haman & Luniewska, 2013). Fr research purpses (e.g. in experimental designs), mnlingual children are regarded as the cntrl grup, but the CLTs can equally well be used simply fr this grup if necessary. Nrming data fr each language still need t be btained befre we can cnsider the CLTs t be a tl fit fr diagnsis. Mrever, during the nrmative data cllectin in varius languages fr the CLTs it will be essential t carefully define the ppulatins fr which a given nrm is t be valid. It was already shwn that the amunt f input (Gathercle et al, 2008) r f cumulative expsure (Unswrth et al, 2011) can differentiate grups f bilinguals in terms f language prficiency. Hence, gathering the nrmative data shuld be accmpanied by careful cntrl f current language input and language histry f the child. T this end, the questinnaires designed within COST Actin IS0804 can be used: the Parents f Bilingual Children Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 225, Questinnaire (PABIQ; Tuller, this vlume) r the Parents f Bilingual Children Questinnaire: Infant & Tddler versin (PABIQ:IT; Gatt et al, this vlume). It wuld nt have been pssible t accmplish the CLTs design withut the jint effrt f numerus members f COST Actin IS0804. In all phases f the CLTs develpment, natinal teams wrked t prvide relevant infrmatin (Phases 1 and 2), t adapt and cnduct the AA study in their language (Phase 3), t select the targets and distractrs fr single language versins f the CLTs, and finally t adapt tlife instructins and all relevant materials. The members f WG3 als served as experts in reviewing the pictures prepared fr the CLTs, which significantly imprved the picture database and in particular its crss-cultural adequacy. These teams shuld thus be regarded as c-authrs f particular language versins f the CLTs. } Acknwledgements The CLTs cnstructin prcess utlined in this chapter wuld nt have been pssible withut the enrmus cllective effrts f members and cllabratrs f COST Actin IS0804 (in particular the Plish team within the Actin). We wuld like t express ur immense gratitude t all f them and in particular t Jakub Szewczyk, Shula Chiat, Frenette Suthwd, Hanne Gram Simnsen and Sharn Armn-Ltem. Our recgnitin als ges t all cmpetent judges in Phase 1, all experts in Phase 2, all thse wh recruited participants fr Phase 3 and all participants. Prgramming and technical supprt fr Phase 3 was prvided by Bartimiej Etenkwski. All pictures were drawn by Justyna Kamykwska, the design artist. Research presented in this chapter was partly financed by the Ministry f Science and Higher Educatin f Pland within the prject 'Cgnitive and language develpment f Plish bilingual children at schl entrance age- risks and pprtunities' (Grant N. 809/N-COST/2010/0), awarded t the Faulty f Psychlgy, University f Warsaw: principal investigatrs: Ewa Haman (University f Warsaw) and Zfia Wdniecka (Jagiellnian University). is. ' Ntes^ (1) " Age f acquisitin f target wrds was btained in a separate study with caregivers assessing at what age their children first prduced the wrd (De Bleser & Kauschke ; 2003).. (2) Picture naming agreement was btained in a separate study with adult native, speakers f German. The prprtin f answers with the target wrd prvided fr the picture shuld be at least 80% t qualify the item as shwing an acceptable level f naming agreement. (3) ^Detailed cmparisns f picture naming differences between nuns and verbs will be presented in a separate publicatin (Haman et al, in preparatin). In this chapter, we present the basic utcme f the study: the prcess f selectin f 158

19 227 Part 2: Phnlgical, and Lexical Prcessing ^ Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 219 nuns and 142 verbs which frm a list f.'clt-candidate wrds shared acrss mst f the languages invlved. (4) Sme ther labels fr this picture were a hspital, a drug stre and a chemist. (5) The standardised length in phnemes was dubled in rder t emphasise the impact f this factr n verall cmplexity f the wrd. (6) Data included in the analysis cmprise Afrikaans, Suth African English, British English, "Catalan, Finnish, German, Hebrew, Italian, Lebanese, Lithuanian, Luxembu'rgish, Maltese, Nrwegian, Plish, Russian, Serbian, Slvak, Swedish and Turkish. References Alari^ F.X. and Ferrand, L. (1999) A set f 400 pictures standardized fr French: Nrms fr name agreement, image agreement, familiarity, visual cmplexity, image variability, and age f acquisitin. Behavir Research Methds 31 (3), Allman, B. (2005) Vcabulary size and accuracy f mnlingual and bilingual preschl children. In J. Chen, K.T. McAlister, K. Rlstad and J. MacSwan (eds) Prceedings - f the 4th Internatinal Sympsium n Bilingualism (pp ). Smerville, MA: Cascadilla Press.!, Andreu, L., Sanz-Trrent, M. and Guardia-Olms, J. (2012) Auditry wrd recgnitin ^ f nuns'and verbs in children with specific language impairment (SLI). Jurnal f 'Cmmunicatin Disrders 45 (1), Armn-Ltem, S. (2011) Intrductin: Bilingual children with SLI - the nature f the prblem-. Bilingualism: Language and Cgnitin 15 (1) ; 1-4. Baayen, R.H., Feldman, L.B. and Schreuder, R. (2006) Mrphlgical influences n the recgnitin f mnsyllabic mnmrphemic wrds. Jurnal f Memry and Language 55 (2), Barca, L., Burani, C. and Arduin, L.S. (2002) Wrd naming times and psychlinguistic nrms fr Italian nuns. Behavir Research Methds 34 (3),, Barnett, W.S. and Lamy, C. (2006) Estimated Impacts f Number f Years f Preschl Attendance n Vcabulary, Literacy and Math Skills at Kindergarten Entry. New Brunswick, NJ: Natinal Institute fr Early Educatin Research. Barry, G, Hirsfy K.W., Jhnstn, R.A and Williams, C.L. (2001) Age f acquisitin, wrd frequency, and the lcus f repetitin priming f picture namin g. Jurnal f Memry and Language 44 (3), Bates, E., D'Amic, S., Jacbsen, T, Szekely, A., Andnva, E., Devescvi, A. and Tzeng, O. (2003) Timed picture naming in seven languages. Psychnmic Bulletin & Review 10 (2), Becker, B. (2011) Scial disparities in children's vcabulary in early childhd. Des preschl educatin help t clse the gap? The British Jurnal f Scilgy 62 (1), Benedict, H. (1979) Early lexical develpment: Cmprehensin and prductin. Jurnal f ^ Child Language 6 (2), Bialystk, E., Craik, F.I.M. and Luk, G. (2008) Lexical access in bilinguals: Effects f vcabulary size and executive cntrl. Jurnal f Neurlinguistics 21 (6), Bialystk, E., Luk, G., Peets, K.F. and Yang, S. (2010) Receptive vcabulary differences in mnlingual and bilingual children. Bilingualism: Language and Cgnitin 13 (4), " > Black, M. and Chiat, S, (2003) Nun-verb dissciatins: A multi-faceted phenmenn. Jurnal f Neurlinguistics 16 (2-3), ! Bnin, P., Perret, C., Met, A., Ferrand, L. and Mermilld, M. (2008) Psychlinguistic nrms and face naming times fr phtgraphs f celebrities in French. Behavir Research Methds 40 (1), ^ } Brdeur, M.B., Dinne-Dstie, E., Mntreuil, T. and Lepage, M. (2010) The Bank f ^ Standardized Stimuli (BOSS), a new set f 480 nrmative phts f bjects t be used as visual stimuli in cgnitive research. PlS ONE 5 (5),el0773. BrWn, G.D. and Watsn, F.Li (1987) First in, first ut: Wrd learning age and spken wrd frequency as predictrs f wrd familiarity and wrd naming latency. Memry & Cgnitin 15 (3), Cameira, M.L. and Vicente, S.G. (2010) Age-f-acquisitin nrms fr a set f 1,749 ^Prtuguese wrds. Behavir Research Methds 42 (2), Capne, N.C. and McGregr, K.K. (2005) The; effect f semantic representatin n tddlers' wrd retrieval. Jurnal f Speech, Language/and Hearing Research 48 (6), Carrll, J.B. ^nd White, M.N. (1973) Age-f-acquisitin nrms fr 220 picturable nuns. Jurnal f Verbal Learning and Verbal Behavir 12 (5), Chen, H.C. (1990) Lexical prcessing in a nn-native language:' Effects f language prficiency and learning strategy. Memry Cgnitin 18 (3), Chiang, P.Y. and Rvachew, S. (2007) English-French bilingual children's phnlgical awareness and vcabulary skills. Canadian Jurnal f Applied Linguistics (CJAL)/Revue canadienne de linguistique appliquee (RCLA) 10 (3), Clark/E.V. (2009) First Language Acquisitin (2nd edn). Cambridge: Cambridge University 'Press., Crtese, M.J. and Khanna, M.M. (2008) Age f acquisitin ratings fr 3,000 mnsyllabic wrds. Behavir Research Methds 40 (3), DAmic, S., Devescvi, A. and Bates, E. (2001) Picture naming and lexical access in Italian children and adults. Jurnal f Cgnitin and Develpment 2 (1), Dale, P.S. "and Penfld, M. (2011) Adaptatins f the MacArthur-Bates I int Nn-U.S. English Languages. See (accessed 12 December 2014). De Bleser, R. and Kauschke, C. (2003) Acquisitin and lss f nuns and verbs: Parallel rdivergent patterns? Jurnal f Neurlinguistics 16 (2), Dijkstra, T. (2003) Lexical prcessing in bilinguals and multilinguals: The wrd selectin 1 prblem. In J. Cenz, B. Hufeisen and U. Jessner (eds) The Multilingual Lexicn (pp ). Netherlands: Springer. Dixn, L.Q. (2011) The rle f hme and schl factrs in predicting English vcabulary amng bilingual kindergarten children in Singapre. Applied Psychlinguistics 32 (1), Dunn, L.M. and Dunn, L.M. (1997) PPVT-III: Peabdy Picture Vcabulary Test. Circle Pines, MN: American Guidance Service. Dunn^L.M., Dunn, L.M., Whettn, C. and Burley, J. (2009) The British Picture Vcabulary Scale (3nd edn). Lndn: GL Assessment. Duursma, E. ; Rmer-Cntreras, S. Szuber, A., Prctr, P., Snw, C. ; August, D. and 1 Cajdern, M. (2007) The rle f hme literacy and language envirnment n bilinguals' English and Spanish vcabulary develpment. Applied Psychlinguistics 28 (1), Elin Tlirdardttir, E. (2011) The relatinship between bilingual expsure and vcabulary develpment. Internatinal Jurnal f Bilingualism 15 (4), Ellis, A.W. and Mrrisn, C.M. (1998) Real age-f-acquisitin effects in lexical retrieval. Jurnal f Experimental Psychlgy: Learning, Memry, and Cgnitin 24 (2), Etenkwski, B. and Luniewska, M. (2012) The Crss-linguistic Lexical Tester (Versin 1.0) [Cmputer sftware] Bartimiej Etenkwski - prjek'twanie i prgramwanie < systemw kmputerwych. EURYDICE at NFER (2010) Cmpulsry age f starting schl in Eurpean cuntries. See (accessed 1 March 2013).

20 228 Part 2: Phnlgical, and Lexical Prcessing ^ Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 219 Fensn, L. ; Dale, P.S., Reznick, J.S. and Thai, D. (1993a) The Mac Arthur-Bates Cmmunicative Develpment Inventries (I I): Wrds and Gestures. Baltimre, MD: Paul Brkes Publishing: Fensn, L., Dale, P.S., Reznick, J.S. and Thai, D. (1993b) The MacArthur-Bates Cmmunicative Develpment Inventries (I II): Wrds and Sentences. Baltimre, MD: Paul Brkes Publishing. Fernald, A c., Perfrs, A. and Marchman, V.A. (2006) Picking up speed in understanding: Speech prcessing efficiency and vcabulary grwth acrss the 2nd year. Develpmental Psychlgy 42 (1), ^ Fernandez, M.C., Pearsn, B.Z., Umbel, V.M., Oiler, D.K. and Mlinet-Mlina, M. (1992) Bilingual receptive vcabulary in Hispanic preschl children. Hispanic jurnal f ' Behaviral Sciences 14 (2), Ferrand, L., Bnin, P., Met, A., Augustinva, M. ; New, B., Pallier, C. and Brysbaert, M..(20Q8) Age-f-acquisitin and subjective frequency estimates fr all generally knwn mnsyllabic French wrds and their relatin with ther psychlinguistic variables. Behavir Research Methds 40 (4), Gagarina, N., Klassert, A. and Tpaj, N. (2010) Sprachstandstest Russisch fur r mehrsprachige Kinder [Russian language prficiency test fr multilingual children]. ; ZAS Papers in Linguistics 54. Gardner, M.F. (1979) Expressive One-Wrd Picture Vcabulary Test. Nvat, CA: Academic Therapy Publicatins. Garlck, V.M., Walley, A.C. and Metsala, J.L. (2001) Age-f-acquisitin, wrd frequency, and neighbrhd density effects n spken wrd recgnitin by children and adults. Jurnal f Memry and Language 45 (3), Gathercle, V.G., Thmas, E.M. and Hughes, E. (2008) Designing a nrmed receptive vcabulary test fr bilingual ppulatins: A mdel frm Welsh. Internatinal Jurnal f Bilingual Educatin and Bilingualism 11 (6), Gatt, D., Letts, C. and Klee, T. (2008) Lexteal mixing in the early prductive vcabularies f Maltese children: Implicatins fr interventin. Clinical Linguistics & Phnetics 22 \ (4-5), Gentner, D. (1981) Sme interesting differences between verbs and nuns. Cgnitin and Brain Thery 4 (2), Gentner, D. (1982) Why nuns are learned befre verbs: Linguistic relativity versus natural partitining. In S. A. Kuczaj (ed.), Language develpment: Vl. 2. Language, thught and culture (pp ). Hillsdale, NJ: Erlbaum. Gentner, D. (2006) Why verbs are hard t learn. In~K. Hirsh-Pasek and R. Glinkff J (eds) Actin Meets Wrd: Hw Children Learn Verbs (pp ). Oxfrd, UK: Oxfrd University Press. Gilhly,-K.J. and Gilhly, M.L.M. (1980) The validity f age-f-acquisitin ratings. ' British Jurnal f Psychlgy 71 (1), , Gilhly, K.J. and Lgie, R.H. (1980) Age-f-acquisitin, imagery, cncreteness, familiarity, and ambiguity measures fr 1,944 wrds. Behavir Research Methds 12 (4), Gldfield, B.A. (200,0) Nuns befre verbs in cmprehensin vs. prductin: The view frm pragmatics. Jurnal f Child Language 27 (3), Gpnik, A., Chi, S. and Baumberger, T. (1996) Crss-linguistic differences in early semantic and cgnitive develpment. Cgnitive Develpment 11 (2), Grman, B/K. (2012) Relatinships between vcabulary size, wrking memry, and phnlgical awareness in Spanish-speaking English language learners. American Jurnal f Speech-Language Pathlgy 21 (2), Gray, S., Plante, E., Vance, R. and Henrichsen, M. (199?) The diagnstic accuracy f fur vcabulary tests administered t preschl-age children. Language, Speech, and Hearing Services in Schls 30 (2), ; Haman, E. (2009) University f Warsaw Picture Database fr Vcabulary Test (Obrazkwy Test Siwnikwy - Rzumienie OTSR). Unpublished material. Faculty c f Psychlgy, University f Warsaw. Haman, E., Luniewska, M. and Pmiechwska, B. (2011, Nvember). Preparatry study fr assessing wrd difficulty: An update n age f acquisitin (AA study). Paper ^ presented at the 5th General Meeting f COST IS0804, Malta. Haman, E. and Frnczyk, K. (2012) Obrazkwy Test Shwnikwy - Rzumienie (OTSR). ^Gdansk: Pracwnia Testw Psychlgicznych i Pedaggicznych. Haman, E. and Luniewska, M. (2013, May) Crss-linguistic lexical tasks (CLTs) assessing swrd knwledge and lexical prcessing in bilingual children. Plenary talk presented.-at Child Language Impairment in Multilingual Cntexts (final COST IS0804 cnference), Krakw, Pland. Haman, E., Szewczyk, J., Luniewska, M., Mieszkwska, K., Pmiechwska, B., Wdniecka, Z., Chiat, S. and Armn-Ltem, S. (in preparatin) Naming bjects and actins acrss 34 languages: Can we establish a cmmn set f nuns and verbs? Hammer, C.S., Lawrence, F.R. and Micci, A.W. (2008) Expsure t English befre and after entry int Head Start 1: Bilingual children's receptive language grwth in Spanish and English. Internatinal Jurnal fbilingual Educatin and Bilingualism 11 (1), Harris, M., Yeeles, C., Chasin, J. and Oakley, Y. (1995) Symmetries and asymmetries in early lexical cmprehensin and prductin. Jurnal f Child Language 22 (1), Hart, B. and Risley, T.R. (2003) The early catastrphe: The 30 millin wrd gap by age 3. \ American Educatr 27 (1), 4-9. Hewitt, L.E., Hammer, C.S., Ynt, K.M. and Tmblin, J.B. (2005) Language sampling fr kindergarten children with and withut SLI: Mean length f utterance, IPSYN, and NI>W. Jurnal f Cmmunicatin Disrders 38 (3), Hff,,(2009) D vcabulary differences explain achievement gaps and can vcabularytargeted interventins clse them? Paper presented at the Natinal Research Cuncil Wrkshp n the Rle f Language in Educatin, Octber. Huebler, F. (2010) Internatinal educatin statistics: Primary schl entrance age and duratin. Internatinal Educatin Statistics. See age.html (accessed 30 May 2010). Jared, D., Pei Yun Ph, R. and Paivi, A. (2012) LI and L2 picture naming in Mandarin- English bilinguals: A test f bilingual dual cding thery. Bilingualism: Language and Cgnitin 16 (2), Juhasz," B.J., Starr, M.S., Inhff, A.W. and Placke, L. (2003) The effects f mrphlgy n the/prcessing f cmpund wrds: Evidence frm naming, lexical decisins and eye fixatins. British Jurnal f Psychlgy 94 (2), Kambanars, M. (2003) Verb and nun prcessing in late bilingual individuals with anmic aphasia. Dctral dissertatin, Flinders University, Adelaide. Kambanais, M. ; Grhmann, K.K. and Thedru, E. (2010) Actin and bject naming in mn- and bilingual children with language impairment. In A. Btinis (ed.) Prceedings f ISCA Tutrial and Research Wrkshp n Experimental Linguistics 2010 (pp ). Kan, P.F. and Khnert, K.J. (2005) Preschlers learning Hmng and English: Lexicalsemantic skills in LI and 12. Jurnal f Speech, Language and Hearing Research 48 (2), Kauschke, C:, Lee, H.W. and Pae, S. (2007) Similarities and variatin in nun and verb acquisitin: A crsslinguistic study f children learning German, Krean, and Turkish; Language and Cgnitive Prcesses 22 (7), Keil, F.C. (1989) Cncepts, Kinds, and Cgnitive Develpment. Cambridge, MA: The MIT - Press.. Khmsi, A. (2001) Evaluatin du Langage Oral. Paris: ECPA. Khnert, KJ.,(2010) Bilingual children with primary language impairment: Issues, evidence and implicatins fr clinical actins. Jurnal f Cmmunicatin Disrders 43 (6),

21 230 Part 2: Phnlgical and Lexical Prcessing V*.- h> ' Khnert, K.J., Hernandez) A. and Bates, E. (1998) Bilingual perfrmance n the Bstn naming test: Preliminary nrms in Spanish and English. Brain and Language 65 (3), Khnert, K.J., Bates, E. and Hernandez, A.E. (1999) Balancing bilinguals: Lexicalsemantic prductin and cgnitive prcessing in children learning Spanish and English. Jurnal f Speech, Language and Hearing Research 42 (6), Khnert, K.J. and Bates, E. (2002) Balancing bilinguals II: Lexical cmprehensin and cgnitive prcessing in children learning Spanish and English. Jurnal f Speech, Language, and Hearing Research 45 (2), Kunnari, S., Savinainen-Makknen, T. and Saarist-Helin, K. (2012) Fnlgiatesti: Lasten aanteellisen kehityksen arviinti [Finnish Test f Phnlgy: Assessment f Phnlgical Skills in Children]. Jyvaskyla: Niil Maki Instituutti. Lahey, M. and Edwards, J. (1996) Why d children with specific language impairment name pictures mre slwly than their peers? Jurnal f Speech and Hearing Research 39 (5)/" Lahey, M., Edwards, J. and Munsn, B. (2001) Is prcessing speed related t severity f language impairment 1Jurnal f Speech, Language, and Hearing Research 44 (6), Lenard, L/B. and Deevy, P. (2004) Lexical deficits in specific language impairment. In L. Verheven and H. van Balkm (eds) Classificatin f Develpmental Language Disrders (pp ). Mahwah, NJ: Lawrence Erlbaum. Lind, M., Simnsen, H.G., Hansen, P., Hlm, E. and Mevik, B.-H. (2013) ^Ordfrradet" - en leksikalsk database ver et utvalg nrske rd. Nrsk tidsskrift fr lgpedi 59 (1), Lind, M., Simnsen, H. G., Hansen, P., Hlm, E. and Mevik, B. H. (2015) Nrwegian Wrds: a lexical database fr clinicians and researchers. Clinical Linguistics and Phnetics. Early nline view (accessed 14 January 2015), Lizu, M. and Stuart, M. (2003) Phnlgical awareness in mnlingual and bilingual English and Greek five-year-lds. Jurnal f Research in Reading 26 (1), Luniewska, M., Pmiechwska, B., Suthwd, F., Slancva, D., Kapalkva, S., Ege, P., Unal, O. and Haman, E. (2012, May) Measuring the age f acquisitin f wrds fr new crssv linguistic lexical tasks: Results f n-line study fr Afrikaans, Plish, Slvakand Turkish. Pster presented at the 6th General Meeting f COST IS0804, Berlin, Germany. Luniewska, M. et al. (2012, September) Age f acquisitin f Bi-Sli WG3 'best wrds' - An update fr 15 languages. Paper presented at the 7th General Meeting f COST IS0804, Padua, Italy. Luniewska, M. and Bi-SLI WG3 (2013, February) AA results fr 15 languages: New data & new analyses. Paper presented at the 8th General Meeting f COST IS0804, Lisbn, Prtugal. Luniewska, M., Andelkvic, D., Armn-Ltem, S., Chiat, S., Dabasinskiene, I., Ege, P.,..., Haman, E. (2014, July). AA nrms fr nuns and verbs in 22 languages. Pster presented at the 13th Internatinal Cngress fr the Study f Child Language, Amsterdam, The Netherlands. Luniewska, M., Haman, E., Armn-Ltem, S., Etenkwski, B., Andelkvic, D., Blm, E.,..., Unal-Lgacev,^ O. (submitted). Ratings f age f acquisitin f 299 wrds acrss 25 languages. Is there a crss-linguistic rder f wrds? Mainela-Arnld, E., Evans, J.L. and Cady, J. A. (2010) Explaining lexical-semantic deficits in specific language impairment: The rle f phnlgical similarity, phnlgical wrking memry, and lexical cmpetitin. Jurnal f Speech, Language, and Hearing Research 53 (6), Mancilla-Martinez, J. v and Lesaux, N.K. (2011) Early hme language use and later vcabulary develpment. Jurnal f Educatinal Psychlgy 103 (3), Marchman, V.A., Fernald, A. and Hurtad, N. (2009) Hw vcabulary size in tw languages relates t efficiency in spken wrd recgnitin by yung Spanishv. English bilinguals. Jurnal f Child Language 37 (4), ^ Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 231 Marques, J.F., Fnseca, F.L., Mrais, S. and Pint, I.A. (2007) Estimated age f acquisitin ^ nrms fr 834 Prtuguese nuns and their relatin with ther psychlinguistic Variables. Behavir Research Methds 39 (3), Mastersn, J. and Druks, J. (1998) Descriptin f a set f 164 nuns and 102 verbs matched fr printed wrd frequency, familiarity and age f acquisitin. Jurnal f - Neurlinguistics 11.(4), Mastersn, J., Druks, J. and Gallienne, D. (2008) Object and actin picture naming in /three- and five-year-ld children. Jurnal f Child Language 35 (2), McDnugh, C., Sng, L., Hirsh-Pasek, K., Glinkff, R.M. and Lannn, R. (2011) An image is wrth a thusand wrds: Why nuns tend t dminate verbs in early wrd learning. Develpmental Science 14 (2), McGregr, R.K., Newman, R.M., Reilly, R.M. and Capne, N.C. (2002) Semantic representatin and naming in children with specific language impairment. Jurnal f Speech, Language and Hearing Research 45 (5), Messer, D. and Dckrell, J.E. (2006) Children's naming and wrd-finding difficulties: Descriptins and.explanatins. Jurnal f Speech, Language and Hearing Research ; 49 (2), " Messer, M.H. (2010) Verbal shrt-term memry and vcabulary develpment in ^mnlingual Dutch and bilingual Turkish-Dutch preschlers. Igitur Archief - Utrecht Publishing and Archiving Service. See Open Educatinal Resurces (OER) prtal at (accessed 10 June 2012). Mntgmery, J.W. (2002) Examining the nature f lexical prcessing in children with specific language impairment: Tempral prcessing r prcessing capacity deficit? Applied Psychlinguistics 23 (3), Mrrisn, C.M., Ellis, A.W. and Quinlan, P.T. (1992) Age f acquisitin, nt wrd frequency, * affects bject naming, nt bject recgnitin. Memry & Cgnitin 20 (6), Mrrisn, C.M., Chappell, TD. and Ellis, A.W. (1997) Age f acquisitin nrms fr a large set f bject names and their relatins t adult estimates and ther variables. The Quarterly Jurnal f Experimental Psychlgy 50A (3), Paradis, J. (2007) Bilingual children with specific language impairment: Theretical and applied issues. Applied Psychlinguistics 28 (3), Pearsn, B.Z., Fernandez, S.C. and Oiler, D.K. (2006) Lexical develpment in bilingual infants. and tddlers: Cmparisn t mnlingual nrms. Language Learning 43 (1), Pena, E.D., Bedre, L.M. and Fiestas, C. (2013) Develpment f bilingual semantic nrms: ^ Gan tw be ne? In V.C.M. Gathercle (ed.) Slutins fr the Assessment f Bilinguals ((pp ). Bristl: Multilingual Matters. Perez,,M.A. and Navaln, C. (2005) Objective-AA nrms fr 175 names in Spanish: Relatinships with ther psychlinguistic variables, estimated AA, and data frm mther languages. Eurpean Jurnal f Cgnitive Psychlgy 17 (2), Pere^ M.A., Izura, C., Stadthagen-Gnzalez, H. and Mann, J. (2013) Assessment f JJ<- bilinguals^ perfrmance in lexical tasks using reactin times. In V.C.M. Gathercle D ^ (ed.) Issues in the Assessment f Bilinguals (pp ). Bristl: Multilingual Matters. Pind, J., Jnsdttir, H., Gissurardttir, H. and Jnssn, F. (2000) Icelandic nrms fr the Sndgrass and Vanderwart (1980) pictures: Name and image agreement, familiarity, and age f acquisitin. Scandinavian Jurnal f Psychlgy 41 (1), Reznick, J.S. and Gldfield, B.A. (1992) Rapid change in lexical develpment in cmprehensin and prductin. Develpmental Psychlgy 28 (3), Rssin, B. and Purtis, G. (2004) Revisiting Sndgrass and Vanderwart ; s bject " pictrial set: The rle f surface detail in basic-level bject recgnitin. Perceptin 33 (2), Santiag, J., MacKay, D.G., Palma, A. and Rh, C. (2000) Sequential activatin prcesses in prducing wrds and syllables: Evidence frm picture naming. Language and "Cgnitive Prcesses 15 (1), 1-44.

22 232 Part 2: Phnlgical and Lexical Prcessing Designing Crss-Linguistic Lexical Tasks (CLTs) fr Bilingual Preschl Children 233 Scheele, A. (2010) Hme language and mn- and bilingual children's emergent academic language. A lngitudinal study f Dutch, Mrccan-Dutch, and Turkish-Dutch 3- t 6-lds. Unpublished dctral dissertatin, University f Amsterdam. Sheng, L. and McGregr, K.K. (2010) Lexical-semantic rganizatin in.children with specific language impairment. Jurnal f Speech, Language, and Hearing Research 53 (1), Simnsen, H.G., Lind, M. and Hansen, P. (2012, May) Cllecting AA data fr Nrwegian: An alternative methd. Paper presented at the 6th General Meeting f COST IS0804, Berlin, Germany. Skipp, A., Windfuhr, K.L. and Cnti-Ramsden, G. (2002) Children's grammatical categries f verb and nun: A cmparative lk at children with specific language impairment (SLI) and nrmal language (NL). Internatinal Jurnal f Language & Cmmunicatin Disrders 37 (3), v Suthwd, R, Luniewska, M., Pmiechwska, B. and Haman, E, (2011, Nvember). Lexical tasks cnstructin: Subjective age f acquisitin f wrds (AA). Data fr Afrikaans and Plish. Pster presented at the 5th General Meeting f COST IS0804, - ; Malta. ~ Spaulding, T.J., Swartwut Szulga, M. and Figuera, C. (2012) Using nrm-referenced "" tests t determine severity f language impairment in children: Discnnect between US plicy makers and test develpers. Language, Speech, and Hearing Services in Schls 43 (2), ' ' Spaulding, T.J., Hsmer, S. and Schechtman, C. (2013) Investigating the interchangeability - and diagnstic utility f the PPVT-III and PPVT-IV fr children with and withut 1 SLI. Internatinal Jurnal f Speech-Language Pathlgy 15 (5), Stadtha gen-gnzalez, H. and Davis, C.J. (2006) The Bristl nrms fr age f acquisitin, f imageability, andfamiliarity. Behavir Research Methds 38 (4), Szekely, A., Jacbsen, T., D'Amic, S.,_ Devescvi, A., Andnva, E., Herrn, D. and Wicha, N. (2004) A new n-line resurce fr psychlinguistic studies. Jurnal f Memry and Language 51 (2), Szekely, A./ D'Amic, S., )evescvi, A., Federmeier, K., Herrn, D., Iyer, G., Jacbsen, T;, Areval, A.L., Vargha, A. and Bates, E. (2005) Timed actin and bject naming. Crtex 41 (1), Tmasell, M., Akhtar, N., Ddsn, K. and Rekau, L. (1997) Differential prductivity in yung children's use f nuns and verbs. Jurnal f Child Language 24 (2), Umbel, V.M., Pearsn, B.Z., Fernandez, M.C. and Oiler, D.K. (1992) Measuring bilingual children's receptive vcabularies. Child Develpment 63 (4), Unswrth, S., Argyri, F., Crnips, L., Hulk, A., Srace, A. and Tsimpli, I. (2011) On the rle f age f nset and input in early child bilingualism in Greek and Dutch. Applied Psychlinguistics 35 (4), Van Tuijl, C., Leseman, P.P.M. and Rispens, J. (2001) Efficacy f an intensive hme-based educatinal interventin prgramme fr 4-t 6-year-ld ethnic minrity children in the Netherlands. Internatinal Jurnal f Behaviral Develpment 25 (2), Vgel, P.M.,and Cmrie, B. (2000) Appraches t the Typlgy f Wrd Classes. Berlin: Walter de Gruyter. Wierzbicka, A. (1988) The Semantics f Grammar. Amsterdam: Jhn Benjamins Publishing Cmpany... Windfufrr, K.L., Faragher, B. and Cnti-Ramsden, G. (2002) Lexical learning skills in yung" children with specific language impairment (SLI). Internatinal Jurnal f Language & Cmmunicatin Disrders 37 (4), ' Zwitserld, P., Blte, J. and Dhmes, P. (2000) Mrphlgical effects n speech prductin: Evidence frm picture naming. Language and Cgnitive Prcesses 15 (4-5), ^ " ". s A Appendix A: The List f CLT-Candidate Wrds (English Equivalents f 158 Nuns and 142 Verbs) Selected in Phase 1 with Values f DNI and MAI Used fr Selectin Nuns Verbs Wrd MAI DNI Wrd MAI DNI airplane bark ant bath apple beg axe bite ball, blw balln bil banana bx barrel break basket brush battery build bear r burn bed A burst bell carry belt carve bicycle clap bird 1.00 ~ 0.96 clean blackbard climb J \ bne cmb bt cnduct bttle ( ck brm crawl brush cry bus\ cut butterfly dance 0.99, 0.99 buttn^ 1.00 J 0.99 dig candle ^ dive cap draw car drill (icntinued)