Evaluation of the Written Competence of Pre-service Teachers of French as a Foreign Language in Belgium

Po a Lingua um 27, ene o 2017 141-153 E alua ion o he W i en Compe ence o P e-se ice Teache s o F ench as a Fo eign Language in Belgium A iane uy eLae Ghen Uni e si y, Belgium Recei ed: 1 Ma ch 2016 / Accep ed: 30 Sep embe 2016 ISSN: 1697-7467 ABSTRACT: The main objec i e o his s udy was o desc ibe F ench as a o eign language (FFL) p o iciency in p e-se ice FFL eache s. This s udy was pe o med wi h all na i e Bel- gian Du ch pos g adua e s uden s (n=40) cou sing an Academic Teache Educa ion P og am (Ghen , Belgium) du ing he academic yea s 2013-2014 and 2014-2015. Analyses e ealed in e es ing indings on hei FFL le el and di e en e o ypes in hei w i en p oduc ions. In conclusion, he p esence o w i ing e o s in FFL in p e-se ice eache s sugges s he need o design no el eaching s a egies o imp o e hei o e all language p o iciency and w i ing skills du ing hei deg ee s udies. Keywo ds: F ench as a o eign language; w i ing e o s; p e-se ice eache s; ©DIALANG; Co puSc ip E aluación de la compe encia esc i a de p o eso es en o mación de ancés como len- gua ex anje a en Bélgica RESUMEN: El obje i o p incipal de es e es udio ue desc ibi las compe encias en ancés como lengua ex anje a (FLE) en u u os p o eso es de FLE. Es e es udio se ealizó con odos los es udian es nee landó onos (n=40) que cu saban el Más e uni e si a io en p o- eso ado de educación secunda ia obliga o ia y bachille a o (Gan e, Bélgica) du an e 2013- 2014 y 2014-2015. Los análisis e ela on esul ados in e esan es sob e su ni el en FLE y di e en es ipos de e o es en sus p oducciones esc i as. En conclusión, la p esencia de e o es esc i os en FLE en p o eso es en o mación sugie e la necesidad de diseña nue as es a egias didác icas pa a mejo a sus compe encias esc i as du an e sus es udios de g ado. Palab as cla e: F ancés como lengua ex anje a; e o es esc i os; p o eso es en o mación; ©DIALANG; Co puSc ip 1. In oduc Ion In he acquisi ion o F ench as a o eign language (FFL), i is a challenge o achie e a nea -na i e le el. This is ce ainly impo an in he case o u u e FFL eache s, esponsible o ans e hei knowledge o enable seconda y, unde g adua e and/o pos g adua e s uden s o mas e i (Myhill, Jones & Wa son, 2013). In o de o each any subjec e ec i ely, i is gene ally accep ed ha an adequa e deg ee o subjec knowledge is equi ed (Mad id, 2004). In addi ion, wi hin he Eu opean Union i is eache s’ linguis ic p o iciency ha o en appea s Po a Lingua um Nº 27, ene o 2017 142 o be he ocus o mos a en ion (T ujillo & Mad id, 2001; Piza o, 2013). Fu he mo e, esea ch sugges ed ha high p o iciency is essen ial o each s uden s o become linguis ically compe en hemsel es (Woodga e-Jones, 2008). In his con ex , s uden s en olled in a Mas e in F ench philology o F ench ansla ion and in e p e a ion s udies should each o a C1 le el o FFL based on Common Eu opean F amewo k o Re e ence’s (CEFR) scales a he end o hei uni e si y s udies. Howe e , a p esen i is s ill unknown i FFL p e-se ice eache s en olled in he Academic Teache Educa ion P og am (ATEP) a Ghen Uni e si y (Belgium) ha e success ully acqui ed his equi ed le el. In his sense, his s udy in ends o shed some ligh on his ques ion. 2. heo e Ical amewo k 2.1. Language p o iciency assessmen In gene al, he CEFR desc ibes o eign language p o iciency a six le els, A1 and A2 (basic use ), B1 and B2 (independen use ), and C1 and C2 (p o icien use ). In his ega d, o eign language (FL) le el can be e alua ed by o icial es s a au ho ized cen e s (e.g. Alliance F ançaise) o h ough online ools and es s. In his sense, ©DIALANG, an online es , allows o assess language p o iciency in 14 di e en Eu opean languages. I has been de eloped by mo e han 20 Eu opean ins i u ions and is based on he Council o Eu ope’s CEFR, which has become es ablished h oughou Eu ope as he mos widely ecognized ame o e e ence in he ield o language lea ning (Zhang & Thompson, 2004; Alde son & Huh a, 2005). In addi ion, i was success ully used o FL le el assessmen in esea ch (Co ina Pé ez, 2011; Lundell & Lindq is , 2014b). 2.2. W i ing skills and e o s In gene al, w i ing has been a use ul ool o assess lea ne s’ FL p o iciency, pa icula ly in academic con ex s (Bene en o & S o ch, 2011; Ma ínez, 2015). In w i ing assignmen s, s uden s a e expec ed o demons a e hei skills and lea ne s pe cei ed ha an ex ensi e ocabula y is c ucial o hei academic w i ing (Leki & Ca son, 1994). Addi ionally, FL w i ing is a complex ac i i y and is conside ed one o he mos di icul skills, due o he combina ion o g amma , ocabula y and discou se o ganiza ion (Llach, 2007; Yu, 2010). Cu en ly, uni e si y p og ams dedica e mo e impo ance o academic w i ing, in eg a ing i mo e in ensi ely and explici ly in he cu iculum (Hade mann & Demeulenae e, 2013). Cu iously, despi e he clea impo ance o w i ing skills in academic con ex s, mos o he s udies in FFL a e mainly ocused on di e en aspec s o g amma (Dewaele & Vé onique, 2000; Bene en o & S o ch, 2011). In addi ion, di e ences in p o iciency be ween na i e speake s and e y ad anced FL lea ne s a e demons a ed. In ac , only a mino i y was capable o acqui e a lexical le el o na i e speake s (Fo sbe g & Ba ning, 2010). Simila - ly, a longi udinal s udy, ocused on he imp o emen o w i ing skills o seconda y school lea ne s o FFL, showed imp o emen s a he discou se le el and in linguis ic complexi y. Howe e , he e we e no signi ican imp o emen s in accu acy and ce ain equen e o s pe sis ed (Bene en o & S o ch, 2011). In addi ion, i was obse ed ha s uden s s ill seem o s uggle wi h aul less academic w i ing (O’Sulli an & Chambe s, 2006). A iAne uy elAe E alua ion o he W i en Compe ence o P e-se ice... 143 Conce ning w i ing e o s, he e a e no uni ied guidelines o hei desc ip ion o as- sessmen . The e o e, linguis s ha e de ised hei own axonomies o e o desc ip ion. Some esea che s used a axonomy based on linguis ic ca ego ies (Ellis, 1994). A mo e ecen and ac ualized classi ica ion o e o s was desc ibed which was no limi ed o g amma ical ele- men s bu also included lexical and s ylis ic issues (James, 2013). James’ classi ica ion se ed as guide o he elabo a ion o he online co ec ion ool called Co puSc ip , de eloped by and cu en ly used a Ghen Uni e si y (Belgium) (Hade mann & Demeulenae e, 2013). This online co ec ion ool dis inguishes h ee main ca ego ies o e o s (lexical, g amma ical and discu si e) and wo o he ca ego ies (spelling mis akes and con en e o s). 3. aIms o he s udy The s udies men ioned abo e demons a ed ha w i ing skills in FL ep esen a g ea challenge o ad anced lea ne s and e en o u u e FL eache s, and i plays a key ole in hei p o essional pe o mance. In his ega d, he p esen s udy was designed based on he impo ance o he acquisi ion o a high FL le el in an academic con ex and he pe sis- ence o w i ing e o s a di e en academic le els. In his con ex , he FFL le el and he equency and ypes o w i en e o s made by na i e Belgian Du ch p e-se ice eache s o FFL a e s ill unknown. In Belgium, all g adua ed u u e FL eache s mus comple e hei o ma ion by ollowing a speci ic pos g adua e ATEP. In his sense, his p og am o e s a unique oppo uni y o e alua e he FL le el and w i ing skills o he u u e FFL eache s. I is hypo hesized ha hese pos g adua e s uden s each a high o close o a na i e le el o FFL. Howe e , p obably hey could s ill commi e o s in hei w i en composi ions which could a ec hei p o essional ca ee . Fo his eason, he aim o his s udy was o de e mine he le el o FFL and o desc ibe quali a i ely and quan i a i ely he ypes and equency o w i en e o s commi ed by hese pos g adua e s uden s. 4. ma e Ial and me hods 4.1. Pa icipan s This s udy was pe o med wi h all he p e-se ice eache s (n=40) en olled in he oblig- a o y ATEP wi h a specializa ion in FFL a Ghen Uni e si y (Belgium) du ing he academic yea s 2013-2014 and 2014-2015. In his popula ion o s uden s, 40 we e na i e Belgian Du ch speake s, which ep esen ed he expe imen al popula ion (EP). EP was composed by 9 male and 31 emale s uden s. In addi ion, he majo i y o he s uden s had acqui ed a Mas e in F ench philology (n=32, PHILG oup and he o he s in F ench ansla ion and in e p e a ion s udies (n=8, TIG oup). No e ha gende in o ma ion was only supplemen a y and i was no used as a iable o o u he analyses in his s udy. 4.2. FFL le el assessmen Du ing he i s s age o his s udy, s uden s we e asked o sel -assess hei FFL le el. In o de o ob ain an objec i e and comple e assessmen o he FFL skills, hey pe o med Po a Lingua um Nº 27, ene o 2017 144 he compu e -based and deli e ed ia he In e ne ©DIALANG es (Zhang & Thompson, 2004) o F ench in an in o ma ics class oom a Ghen Uni e si y. Be o e s a ing he ac ual language skills es , hey pe o med a ocabula y size placemen es , in which a collec ion o wo ds – eal and in en ed e bs– a e p esen ed, and a sel -assessmen ques ionnai e in o de o a e hei own language abili ies (Ockey, 2009). I was sco ed be ween 0-1000 poin s (Zhang & Thompson, 2004). Nex , lis ening, w i ing, eading, g amma and ocabula y skills we e objec i ely e alua ed and sco ed on he CEFR scales. These alues we e hen analyzed o know he o e all language p o iciency in FFL in each s uden . In addi ion, ©DIALANG does no gi e a global sco e, bu sco es ange om A1-C2 ( o al o six le els) in each skill. In his sense, o ob ain a nume ic alue co esponding he global FFL le el o each pa ic- ipan , a numbe om 1-6 was assigned o each le el (1=A1, 2=A2 ill 6=C2 espec i ely). 4.3. W i ing assignmen Du ing he second s age o he s udy, in o de o assess he w i ing skills, s uden s we e asked o pe o m a w i ing assignmen in F ench which was conduc ed in hei own class oom unde hei eache ’s and he esea che ’s supe ision. I consis ed o a sho essay which should include a b ie in oduc ion and a clea conclusion (375-400 wo ds) ela ing o an aspec o FFL and i s eaching. The i le o he essay gi en o he s uden s was: “L’impo - ance de l’enseignemen du ançais langue é angè e dans une Eu ope mul ilingue.” This gene al opic was selec ed hanks o i s di ec ela ion o he s uden ’s backg ound. Thus, speci ic knowledge on he subjec has played no ole. S uden s we e gi en an in oduc ion o 10 minu es ou lining wha hey had o do; hey we e allowed up o 60 minu es o comple e his ask, o e ise hei handw i en ex and make any changes hey wished. Addi ionally, s uden s could no make use o any help sou ce (dic iona y o g amma , no we e hey allowed o ask he eache o esea che o help). 4.4. W i ing e o analysis In he hi d s age o his s udy, all essays we e ea ed in di e en phases. In he i s phase, in o de o acili a e he w i ing e o analysis, he 40 handw i en essays we e collec ed and ansc ibed li e ally in o wo d iles using Mic oso O ice Wo d 2007. In he second phase, in o de o pe o m an e icien , uni o m and objec i e w i ing e o analysis, all essays we e uploaded o he online co ec ion ool Co puSc ip . This ool has been de eloped by Ghen Uni e si y because he e was a g owing need o a mo e e - icien , uni o m and, hence, objec i e co ec ion me hod o w i en assignmen s (Hade mann & Demeulenae e, 2013). Th ee main ca ego ies o e o s a e dis inguished: lexical, g amma - ical and discu si e e o s. Two o he ca ego ies, spelling mis akes and con en e o s, we e also included. In addi ion, he ollowing sub ypes o e o s we e conside ed in he h ee main ca ego ies: omission, misselec ion, o e inclusion and o de (James, 2013). Omission is conside ed as he absence o an i em ha should appea in a well- o med sen ence (e.g. g amma ical omission: *Dans un p emie emps nous nous in é essons à les sui es posi i es pou ini pa (…).); misselec ion is he use o he w ong o m o he mo pheme o s uc u e (e.g. lexical misselec ion: *Ap ès une jou née occupée, (…), ça nous ai du bien (...) pou se elaxe .); o e inclusion is de ined as he p esence o an i em ha should no appea in a A iAne uy elAe E alua ion o he W i en Compe ence o P e-se ice... 145 well- o med sen ence (e.g. g amma ical o e inclusion: *Le con ac physique a ec un li e, (…), es un p emiè e aspec posi i .); and inally o de is ega ded as he inco ec placemen o a mo pheme in an u e ance (e.g. g amma ical o de : *D’ailleu s, ne pas ou le monde joui d’un o dina eu .). Thus, in his s udy, a o al o 14 ypes o w i en e o s we e e alua ed. In he hi d phase o he analysis, in o de o e alua e he equency and ypes o e - o s, all he ansc ibed w i en composi ions we e ead and sc u inized one by one. Fi s , he o al numbe o wo ds w i en in each essay has been calcula ed. Fu he mo e, e o s we e iden i ied, selec ed and classi ied in o he main ca ego ies and sub ypes as desc ibed abo e, and hen coun ed up. Nex , he numbe o o al e o s and he numbe o e o s o each ca ego y and sub ype was w i en down in Excel o each s uden . Finally, he mean equency o each e o ype was calcula ed. A his s age, no co ec ion o eedback was p o ided. Ins ead, he s uden s we e in o med ha hey would ecei e a pe sonal eedback on hei essay and ha hey would ha e he oppo uni y o imp o e hei assignmen . 4.5. Quan i a i e and s a is ical analyses In his s udy, all da a om he FFL le el analysis (©DIALANG) and he w i ing e o analysis we e collec ed and subjec ed o he Shapi o-Wilk es o no mali y by using he so wa e SPSS 15.0. Sco es o he ocabula y placemen es we e no mally dis ibu ed and he e o e, s uden ’s - es was used o de e mine s a is ical di e ences. FFL skills om ©DIALANG and da a om w i ing e o s analyses we e non-no mally dis ibu ed, and Mann-Whi ney non-pa ame ic es was used o de e mine s a is ical signi icance. In addi ion, he esul s o each a iable we e exp essed as mean and s anda d de ia ion (SD) alues and p alues p<0.05 we e conside ed s a is ically signi ican in wo- ailed es s. In w i ing e o s analyses, he pe cen age o s uden s who commi ed each pa icula e o , he dis ibu ion o each ca ego y o w i ing e o s and hei espec i e sub ypes we e p esen ed as pe cen age in hei espec i e ables. 5. esul s 5.1. Le el o FFL Sel -assessmen o he FFL le el pe o med by each s uden anged om B1 o C2 o EP. Fu he mo e, 82.5% o EP placed hemsel es a he C le el on he CEFR scales (27.5% a C1 and 55% a C2) and 17.5% supposed ha ing a B le el (2.5% a B1 and 15% a B2). These esul s sugges ed ha , as a as sel -assessmen is conce ned, EP supposed a high FFL le el, compa able o na i e speake s. A e sel -assessmen , s uden s pe o med he ©DIALANG in o de o ha e a mo e objec i e o e iew on hei FL le el. Fi s , on he ocabula y placemen es , EP sco ed a mean o 759/1000. In e es ingly, esul s by g oup e ealed ha PHILG oup ob ained a highe mean sco e (766/1000) han TIG oup (731/1000) bu no s a is ical di e ences we e ound (p=0.526). Conce ning he acqui ed CEFR le els, esul s by EP showed ha 62.5% was placed a B2 and 37.5% a C1. Obse ing esul s by g oup, 37.5% o bo h g oups we e placed a C1. Po a Lingua um Nº 27, ene o 2017 146 In addi ion, mo e han hal o he pa icipan s (62.5% o PHILG oup and TIG oup) we e placed a a B2 le el. None o hem was placed a C2. Table 1 ©DIALANG analysis o FFL le el by skills ©DIALANG FFL SKILLS G oups Lis ening W i ing Reading G amma Vocabula y EP 4.5 ± 0.6 3.9 ± 0.5 5.0 ± 0.8 4.2 ± 0.7 4.1 ± 0.7 C1 B2 C1 B2 B2 PHIL 4.5 ± 0.7 3.9 ± 0.6 4.9 ± 0.8 4.1 ± 0.6 4.0 ± 0.7 C1 B2 C1 B2 B2 TI 4.4 ± 0.5 4.0 ± 0.5 5.4 ± 0.9 4.5 ± 0.9 4.1 ± 0.6 B2 B2 C1 C1 B2 STATISTICAL ANALYSIS PHIL s. TI p=0.629 p=0.557 p=0.179 p=0.146 p=0.663 FFL esul s a e p esen ed as mean ± s anda d de ia ion alues o each g oup and skills. FFL le el is exp essed acco ding o he CEFR scale. S a is ical p alues o Mann-Whi ney non-pa ame ic es a e shown and all p alues below 0.05 we e conside ed s a is ically signi ican o he wo- ailed es . In ela ion o ©DIALANG analysis o he CEFR le els o each skill (Table 1), EP showed a B2 le el o w i ing, g amma and ocabula y, and a C1 le el o lis ening and eading. Simila ly, PHILG oup ob ained a B2 le el o each skill. Howe e , o ecep i e skills such as lis ening and eading skills, s uden s ob ained a highe le el (C1). Whils , obse ing he sco es ob ained by g oup, means o e e y skill we e e y simila o each g oup. Howe e , o eading skills and g amma , TIG oup ou pe o med PHILG oup. Fi- nally, when PHIL and TIG oups we e compa ed, no s a is ical signi ican di e ences we e obse ed (p>0.05). 5.2. Analysis o w i en composi ions and gene al classi ica ion o he w i ing e o s A sys em o classi ica ion o e o s was based on he axonomy es ablished by James (James, 2013). A summa y o he numbe and pe cen age o e o s made in each main ca - ego y o e o is summa ised in Table 2. The o al analyzed co pus con ained a o al o 11069 wo ds w i en by 40 s uden s. Tex leng h o he essays o he EP anged om 197 o 432 wo ds wi h a mean o 276.73 wo ds, which was unde he es ablished maximum leng h (400 wo ds). Mean leng h o he essays o PHILG oup (271.28 wo ds) was sligh ly lowe han ha o he TIG oup (298.50 wo ds) bu bo h we e s ill unde he maximum es ablished leng h. In addi ion, he sho es and he longes ex we e bo h w i en by s uden s o PHILG oup (Table 2). A iAne uy elAe E alua ion o he W i en Compe ence o P e-se ice... 147 The gene al analysis o e o s e ealed a o al o 163 e o s (100%) in he analyzed co pus. 80% o hese e o s we e commi ed by PHILG oup (131 e o s) and 20% by TIG oup (32 e o s). Rega ding o he o al numbe o e o s, he e we e no signi ican di e ences (p=0.919) be ween bo h g oups (Table 2). Table 2 Gene al analysis o w i ing e o s. MAIN CATEGORIES OF ERRORS % CATEGORY ERROR OF TOTAL ERRORS EP (n=40) PHILG oup (n=32) TIG oup (n=8) LEXICAL 41.1% (67) 75% s uden s 1.68 ± 1.33 75% s uden s 1.69 ± 1.38 75% s uden s 1.63 ± 1.19 GRAMMATICAL 32.5% (53) 70% s uden s 1.33 ± 1.21 68.8% s uden s 1.28 ± 1.11 75% s uden s 1.50 ± 1.60 SPELLING 24.5% (40) 62.5% s uden s 1.00 ± 0.99 62.5% s uden s 1.06 ± 1.05 62.5% s uden s 0.75 ± 0.71 DISCURSIVE 1.2% (2) 5% s uden s 0.05 ± 0.22 3.1% s uden s 0.03 ± 0.18 12.5% s uden s 0.13 ± 0.35 CONTENT 0.61% (1) 2.5% s uden s 0.03 ± 0.16 3.1% s uden s 0.03 ± 0.18 0% s uden s 0.00 ± 0.00 TOTAL ERRORS 100% (163) 100% s uden s (163) 4.08 ± 2.28 100% s uden s (131) 4.09 ± 2.20 100% s uden s (32) 4.00 ± 2.73 TOTAL WORDS CORPUS 11069 11069 276.73 ± 60.52 8681 271.28 ± 59.61 2388 298.50 ± 63.19 The gene al dis ibu ion o each e o ca ego y is p esen ed as pe cen age and numbe o e o s be ween () o each one. In he analysis by g oup, pa icipan s who commi ed each ype o e o a e shown in pe cen - age (% s uden s). W i ing e o s a e shown as mean ± s anda d de ia ion o each e o ca ego y and g oup. O e all analysis o each ca ego y o e o e ealed ha ocabula y e o s we e he mos equen wi h 41.1%, ollowed by g amma ical (32.5%) and spelling (24.5%) e o s. Discu si e and con en e o s ep esen ed 1.2% and 0.61% espec i ely o all e o s (Table 2). Fi s , he analysis o ocabula y e o s showed ha a la ge pe cen age o s uden s (75%) commi ed his ype o e o in each g oup. The mean o ocabula y e o s was highe in PHILG oup bu di e ences be ween g oups we e no s a is ically signi ican (p>0.05). Second, he analysis o g amma ical e o s showed simila esul s whe e 75% o TIG oup commi ed his ype o e o , ollowed by PHILG oup (68.75%). In addi ion, TIG oup showed highe mean o g amma ical e o s bu his di e ence was no s a is ically signi ican (p>0.05). Thi d, he analysis o spelling e o s e ealed ha mo e han hal o he s uden s (62.5%) commi ed Po a Lingua um Nº 27, ene o 2017 148 his ype o e o in each g oup, bu spelling e o s we e less equen han ocabula y and g amma ical e o s. Finally, discu si e and con en e o s we e less equen han all he o he e o s desc ibed abo e and no s a is ical di e ences we e obse ed (Table 2). 5.3. Analysis o sub ypes o w i ing e o s The analysis o syb ypes o w i ing e o s e ealed ha main pa o he sub ypes o e o s was p esen in he h ee main ca ego ies o e o s, especially in g amma ical and lexical e o s, bu clea di e ences we e obse ed in hei dis ibu ion along g oups (Table 3). Fi s , when g amma ical e o s we e deeply analyzed, i was obse ed ha a high pe cen age o s uden s commi ed g amma ical omission and o e inclusion e o s, ollowed by misselec ion and o de . This o de was as well obse ed in he mean alues o e o s o each sub ype (Table 3). Indeed, mean alues in o e inclusion e o s we e he highes in TIG oup, bu wi hou s a is ically signi ican di e ences compa ing o PHILG oup. Omission g amma ical e o s showed he second highes mean alues in TIG oup and did no p esen s a is ical di e ences (Table 3). In he case o g amma ical misselec ion and o de e o s, mean alues we e lowe and no s a is ical di e ences we e iden i ied be ween g oups. Table 3 De ailed analysis o he sub ypes o w i ing e o s. 7  GROUPS SUBTYPES OF WRITING ERRORS DISTRIBUTION OF SUBTYPES OF ERRORS OMISSION MISSELECTION OVERINCLUSION ORDER LEXICAL EP 7.5% s uden s 0.08 ± 0.27 75% s uden s 1.58 ± 1.26 0% s uden s 0.00 ± 0.00 2.5% s uden s 0.03 ± 0.16 MISSELECTION: 94% / 38.7% OMISSION: 4.5% / 1.8% ORDER: 0% / 0% OVERINCLUSION: 1.5% / 0.6% PHIL 6.3% s uden s 0.06 ± 0.25 75% s uden s 1.59 ± 1.32 0% s uden s 0.00 ± 0.00 3.1% s uden s 0.03 ± 0.18 TI 12.5% s uden s 0.13 ± 0.35 75% s uden s 1.50 ± 1.07 0% s uden s 0.00 ± 0.00 0% s uden s 0.00 ± 0.00 GRAMMATICAL EP 47.5% s uden s 0.55 ± 0.64 27.5% s uden s 0.30 ± 0.52 30% s uden s 0.43 ± 0.78 5% s uden s 0.05 ± 0.22 OMISSION: 41.5% / 3.5% OVERINCLUSION: 32.1% / 10.5% MISSELECTION: 22.6% / 7.4% ORDER: 3.8% / 1.2% PHIL 50% s uden s 0.56 ± 0.62 31,3% s uden s 0.34 ± 0.55 25% s uden s 0.34 ± 0.70 3.1% s uden s 0.03 ± 0.18 TI 37.5% s uden s 0.50 ± 0.76 12,5% s uden s 0.13 ± 0.35 50% s uden s 0.75 ± 1.04 12.5% s uden s 0.13 ± 0.35 DISCURSIVE EP 2.5% s uden s 0.03 ± 0.16 2.5% s uden s 0.03 ± 0.16 0% s uden s 0.00 ± 0.00 2.5% s uden s 0.00 ± 0.00 OMISSION: 50% / 0.6% MISSELECTION: 50% / 0.6% OVERINCLUSION: 0% / 0% ORDER: 0% / 0% PHIL 0% s uden s 0.0 ± 0.00 (α; p=0.046) 3.1% s uden s 0.03 ± 0.18 % s uden s 0.00 ± 0.00 3.1% s uden s 0.00 ± 0.00 TI 12.5% s uden s 0.13 ± 0.35 % s uden s 0.00 ± 0.00 0% s uden s 0.00 ± 0.00 0% s uden s 0.00 ± 0.00 Table 3 De ailed analysis o he sub ypes o w i ing e o s The dis ibu ion o each sub ype o e o in each main ca ego y is shown in dec easing o de as he pe cen age o each sub ype o he o al numbe o e o s in i s ca ego y and he pe cen age o each sub ype o he o al e o s p esen in he co pus. In he analysis by g oup, pa icipan s who commi ed each ype o e o a e shown in pe cen age (% s uden s). W i ing e o s in each g oup a e shown as mean ± s anda d de ia ion. (α) shows signi ican di e ences in compa ison o TIG oup and hei espec i e p alue o Mann-Whi ney non-pa ame ic es . All p alues below 0.05 we e conside ed s a is ically signi ican o he wo- ailed es . The dis ibu ion o each sub ype o e o in each main ca ego y is shown in dec easing o de as he pe cen - age o each sub ype o he o al numbe o e o s in i s ca ego y and he pe cen age o each sub ype o he o al e o s p esen in he co pus. In he analysis by g oup, pa icipan s who commi ed each ype o e o a e shown in pe cen age (% s uden s). W i ing e o s in each g oup a e shown as mean ± s anda d de ia ion. (α) shows signi ican di e ences in compa ison o TIG oup and hei espec i e p alue o Mann-Whi ney non-pa ame ic es . All p alues below 0.05 we e conside ed s a is ically signi ican o he wo- ailed es . Second, analysis o lexical e o s showed ha a high pe cen age o s uden s commi ed lexical misselec ion e o s as compa ed o he o he sub ypes o lexical e o s. Fu he mo e, highe bu no signi ican (p>0.05) mean alues we e obse ed in PHILG oup in compa ison o TIG oup. Mean alues o he o he lexical sub ypes w i ing e o s we e conside ably lowe (all below 0.13 e o s) han misselec ion (Table 3). A iAne uy elAe E alua ion o he W i en Compe ence o P e-se ice... 149 Thi d, in-dep h analysis o discu si e e o s con i med he lowe p esence o hese e o s in EP. The only sub ypes o discu si e e o s p esen we e omission ollowed by misselec ion and no s a is ical di e ences we e obse ed, excep o omission (p=0.046). Rema kably, TIG oup only commi ed omission e o s (Table 3). Finally, when he dis ibu ion o sub ypes o e o s was analyzed in ela ion wi h he o al numbe o e o s in i s ca ego y and wi h he o al numbe o e o s in he co pus, a majo a ia ion was obse ed. In he case o g amma ical e o s, he dis ibu ion o sub ypes o e o s was he e ogeneous whe e he mos equen sub ype o e o was omission (41.5%), ollowed by o e inclusion, misselec ion and o de . In e es ingly, g amma ical omission e o ep esen ed 13.5% o all he e o s p esen in he co pus. Unlike g amma ical e o s, in he case o lexical e o s, he sub ypes o e o s we e less dis ibu ed whe e 94% was lexical misselec ion. In con as wi h g amma ical omission e o s, lexical misselec ion e o s ep- esen ed 38.7% o all e o s. A las , he mos equen sub ypes o discu si e e o s we e misselec ion and o de wi h 50% each. Howe e , discu si e misselec ion and omission e o s ep esen ed less han 1% o all e o s. Examples o all he ypes o e o s can be ound in able 4. Table 4 Examples o w i ing e o s 9  EXAMPLES OF WRITING ERRORS OMISSION MISSELECTION OVERINCLUSION ORDER GRAMMATICAL B e , l’app en issage du ançais s’a è e ès u ile de pou oi communique , e (…) la communica ion a ec les pays plus loin ain. B e , si ou Eu opéen doi é udie deux langues, nous dési e ons qu’il de ai ê e l’anglais e le ançais. Posséde des compé ences ançaises, à quoi ça se -il? Ils le eulen su ou app end e pou pou oi a aille , oyage , e c., pa ce que c’es une langue p a ique dans le monde. LEXICAL En ce qui conce ne le ançais, connaî e ce e langue signi ie en même econnaî e la g ande pa ie des Eu opéens qui la pa len . En in, la F ance es le pays le plus ado é comme pays de acance. / Si ous su un de os oyages ai es un e o en e s les gens, ces de nie s e on sans dou e la même chose pou ous. DISCURSIVE / A cause de la p ésence é endue de ce e langue, dans beaucoup de pays, il es nécessai e ou au moins ès u ile pou les non- ancophones d’en a oi une bonne connaissance (…). / / SPELLING La communica ion en Eu ope pou ai s’ammélio e si le ançais ece ai plus d’impo ance à l’école. CONTENT Dans l’Eu ope, les pays les plus impo an s du poin de ue économique son l’Allemagne, les E a s-Unis, mais aussi la F ance ai pa ie de ce g oupe. Table 4 Examples o w i ing e o s Examples o w i ing e o s obse ed in he co pus. E o s a e indica ed in bold. Examples o w i ing e o s obse ed in he co pus. E o s a e indica ed in bold. 6. dIscussIon In he p esen s udy, due o he g ea impo ance o a high FL le el in p e-se ice eache s one o he aims was o de e mine he FFL le el and o assess he w i ing skills o na i e Belgian Du ch pos g adua e s uden s en olled in an ATEP. In his ega d, in o de o de e mine he FFL le el, a combina ion o sel -assessmen (subjec i e) and ©DIALANG (objec i e) we e pe o med and compa ed. In he case o ©DIALANG, i was chosen hanks o i s demons a ed high e icacy, he eliabili y o esul s, and because i allows o pe o m a comple e e alua ion o i e FL skills (Zhang & Thompson, 2004; Ockey, 2009; Klimo a & Hubacko a, 2013). I was a use ul ool bu i was no possible o assess he o al p oduc ion o he s uden s, and he e o e he o e all FFL le el is s ill unknown.