1
A longi udinal s udy o he e ec s o model ex s on EFL child en’s w i en p oduc ion
Ma ía Luquin and Ma ía del Pila Ga cía Mayo
Abs ac
As w i en co ec i e eedback ools, model ex s ha e been claimed o imp o e language
lea ne s’ subsequen p oduc ion, bu almos exclusi ely in e ms o lexical gains. Howe e ,
sca ce esea ch has been ca ied ou wi h EFL child en, an unde ep esen ed popula ion in he
li e a u e, and much less om a longi udinal pe spec i e. The main aim o his s udy was o
de e mine he ex en o which sus ained exposu e o models can ha e an impac on he w i en
p oduc ion o child EFL lea ne s. Thi y pai s o 11–12-yea -old Spanish EFL child en we e
andomly assigned o a con ol g oup, a ea men g oup and a long- e m ea men g oup, who
engaged in wo ou -s age collabo a i e w i ing cycles o h ee weeks each. The child en’s
collabo a i e ex s we e ansc ibed and analyzed conside ing di e en measu es ( ype o
clause, syn ac ic complexi y, lexical di e si y, accu acy, luency and holis ic assessmen ). Ou
indings e ealed ha model ex s led o a dec ease in he numbe o p e-clauses and an inc ease
in he syn ac ic complexi y o he ex s in he sho un. A long exposu e o models showed ha
he child en we e able o p oduce ewe p o o-clauses and mo e clauses, ea u e highe lexical
di e si y in hei ex s and make ewe e o s.
Keywo ds: child en, CAF, EFL, longi udinal design, model ex , longi udinal design
1. In oduc ion
This is he accep ed manusc ip o he a icle ha appea ed in inal o m in Sys em 120 : (2024) // A icle ID 103190, which has been
published in inal o m a h ps://doi.o g/10.1016/j.sys em.2023.103190. © 2023 Else ie unde CC BY-NC-ND license (h p://
c ea i ecommons.o g/licenses/by-nc-nd/4.0/)
2
The wo ldwide in oduc ion o a o eign language (FL) a an ea ly age esponds bo h o a e y
clea social demand and o he con ic ion ha young child en ha e g ea e audi o y and o al
plas ici y, which will allow hem o assimila e he FL be e han a a mo e ad anced age
(Nicholas & Ligh bown, 2008). Howe e , he e is no de ini i e e idence suppo ing he
po en ial ad an ages o ea ly-age FL lea ning (Copland & Ga on, 2014). This lack o cla i y
eme ges as a esul o mul iple ac o s ha hinde he p ocess o language acquisi ion. These
ac o s encompass challenges like o e c owded class ooms, insu icien imme sion in he
a ge language, limi ed chances o in e ac ion ou side he class oom se ing, and a cons ained
cu iculum du a ion (Ga cía Mayo, 2018). I is, he e o e, also impo an o acknowledge he
necessi y o enhancing he amoun o ime and ac i e in ol emen wi h he FL (Manchón,
2009). I is ue ha young lea ne s (YLs) enjoy g ea acili y o he de elopmen o basic
communica ion skills, such as o al comp ehension and p oduc ion, undamen ally linked o
social skills (Kellogg, 2008). In he Spanish con ex , howe e , ou o he ou basic skills in
he FL cu iculum, he w i ing skill has been neglec ed in a o , almos exclusi ely, o he
speaking skill (Cáno as Gui ao, 2017). This lack o emphasis on w i ing clashes wi h esea ch
in he ield o second language acquisi ion (SLA) o ien ed owa d a w i ing- o-lea n
pe spec i e (Manchón, 2011), which sugges s ha lea ne s can bene i om ecei ing
in o ma ion abou he accu acy and app op ia eness o hei ex s, suppo ing he ad ancemen
o hei L2 knowledge (Fe is, 2010).
One way o achie ing his goal is h ough eedback p o ision, which seems o ac i a e
cogni i e p ocesses such as hypo hesis o ma ion and es ing, a en ion, me alinguis ic
e lec ion, no icing (Schmid , 2001), o p oblem-sol ing s a egies (Ga cía He nández e al.,
2017), all o hem deemed c ucial o language acquisi ion. As a esul o i s language
acquisi ion po en ial, w i en co ec i e eedback (WCF) has been ex ensi ely esea ched in
ecen decades (Bi chene & Knoch, 2009). In pa icula , esea che s ha e p oposed model
3
ex s as an e ec i e WCF echnique on he g ounds ha hey cons i u e a mo e discu si e
eedback me hod which does no explici ly single ou e o s bu ea s he ex as a whole.
S udies using models as a o m o WCF ha e been conduc ed bo h wi h adul /adolescen (e.g.,
Hanaoka & Izumi, 2012; Kang, 2020; Ma ínez Es eban & Roca de La ios, 2010; Mon ealeg e
Ramón, 2019; Wu e al., 2023; Yang & Zhang, 2010) and child (e.g., Cáno as Gui ao e al.,
2015; Coyle & Roca de La ios, 2020; C iado e al., 2022; Láza o-Iba ola, 2021; Láza o-
Iba ola & Villa eal, 2021; Roo hoo e al., 2022; Villa eal & Láza o-Iba ola, 2022)
pa icipan s, bu mo e wo k is needed o elucida e whe he he bene i s epo ed in hese s udies
only lead o g ea e p ecision in L2 w i ing o o language de elopmen in he long e m (Polio,
2012). Pedagogically, many EFL eache s seem no o be awa e o he impo ance o WCF so
esea che s should y o ans e knowledge om empi ical s udies o i o play an impo an
ole in language lea ning cu icula. Thus, he main aim o his s udy was o de e mine he ex en
o which sus ained exposu e o models can ha e an impac on he w i en p oduc ion o child
EFL lea ne s.
2. Li e a u e e iew
2.1. Child L2 acquisi ion in o eign language con ex s
Child L2 acquisi ion in FL con ex s is a apidly g owing ield o s udy, as mo e and mo e
schools a ound he wo ld a e in oducing English a he p e-p ima y and p ima y le els (Ga cía
Mayo, 2018). The li e a u e has p esen ed se e al easons o he pe cei ed "has iness" in FL
lea ning. One p ima y a gumen is ad oca ed by go e nmen s, pa en s, and socie y,
emphasizing he need o lea ne s o be p o icien in he FL o access in e na ional educa ion
and employmen oppo uni ies in oday's globalized wo ld (Copland & Ga on, 2014; Ene e
& Moon, 2009; Ga cía Mayo, 2018). Ano he a ionale s ems om he belie ha younge is
4
be e , suppo ed by signi ican bene i s obse ed in imme sion and bilingual se ings (Lys e ,
2007).
Howe e , he po en ial ad an ages o ea ly FL lea ning lack conclusi e e idence (Copland &
Ga on, 2014; DeKeyse , 2013). Va ious ac o s impede language lea ning, such as la ge class
sizes, limi ed exposu e o he a ge language, es ic ed oppo uni ies o in e ac ion ou side
he class oom, and limi ed cu iculum ime (Ga cía Mayo, 2018; Ga cía Mayo & Hidalgo,
2017; Huang, 2015). Age, oo, has been a signi ican a iable in child L2 acquisi ion esea ch,
wi h indings showing ha he no ion o " he younge , he be e " does no uni e sally apply
(Ga cía Mayo & Ga cía Lecumbe i, 2003; Muñoz, 2006).
Conside ing all aspec s, he no ion de i ed om bilingual o imme sion con ex s ha YLs
e o lessly abso b he FL does no apply when ans e ed o en i onmen s wi h limi ed
exposu e o language inpu , especially in la ge g oup se ings (Copland & Ga on, 2014). While
in oducing a FL in p ima y cu iculum may ha e e en ual bene i s o child en, he e is a
no able lack o esea ch on how child en ac ually app oach speci ic lea ning asks in such
con ex s. Fu he mo e, he e is a need o explo e e ec i e eaching me hods ailo ed o YLs'
speci ic needs in hese cons ained se ings compa ed o adul s (Ga cía Mayo, 2018).
One pa icula lea ning ask ha demands u he explo a ion is w i ing, along wi h he
eedback p o ided on he w i en ou pu , which appea s o engage cogni i e p ocesses such as
o mula ing and es ing hypo heses, ocusing a en ion, engaging in me alinguis ic e lec ion,
ecognizing linguis ic pa e ns, and employing p oblem-sol ing s a egies (Williams, 2012). In
English as a Second Language (ESL) and English as a o eign language (EFL) se ings, s uden s
migh be accus omed o ecei ing and assessing eedback, pa icula ly when i comes o w i ing
5
assignmen s. Howe e , his pa e n is mo e p e alen among olde lea ne s who ypically
display g ea e ecep i eness o eedback conce ning hei linguis ic mis akes. Con e sely, he
endency wi h younge lea ne s is o p io i ize o al asks o e w i en asks and eedback
p o ision on hei w i en wo k is o en o e looked. This challenge unde sco es he eason why
child en equen ly lack amilia i y wi h composing in a second o o eign language (Cáno as
Gui ao, 2017). Some au ho s ha e linked his lack o engagemen wi h w i ing o an
unde es ima ion o YLs cogni i e o linguis ic capaci ies (e.g., Láza o-Iba ola, 2023; Muñoz,
2017; Tellie & Roeh -B ackin, 2017; Villa eal & Muna iz-Iba ola, 2021; Villa eal &
Ma ínez-Sánchez, 2023).
The po en ial impac o WCF on acili a ing language acquisi ion has been ex ensi ely
in es iga ed in ecen yea s (Bi chene & Knoch, 2009). I is wi hin his con ex ha explo ing
he con ibu ion o WCF o he language lea ning o child en eme ges as a p omising a enue
o in es iga ion. We also need o know mo e abou he ole ha di e en ypes o eedback
play in L2 lea ning (Ellis, 2008) and he ex en o which o al in e ac ion a ec s w i en
p oduc ion (S o ch, 2021). Addi ionally, ano he big challenge o he SLA ield is he design
o longi udinal s udies, since hey may p o ide aluable insigh s in o changes ha child en
unde go o e ime (Long, 2014).
2.2. WCF in he L2 class oom: model ex s as a w i en eedback al e na i e
SLA esea ch has paid a en ion o he po en ial o eedback in w i en p oduc ion and i s impac
on he de elopmen o he lea ne 's in e language h ough hypo heses es ing (Polio, 2012).
The a gumen is ha eedback is ela ed o he p ocesses o a en ion o o m, which is c ucial
o acquisi ion (Ellis, 2016). Howe e , esea ch on eedback has posed a se ies o limi a ions
6
as o i s e ec i eness when i is deli e ed in he o m o co ec ion o e o s. Al hough posi i e
esul s ha e been epo ed ega ding he e ec s o eedback on a limi ed numbe o linguis ic
o ms (e.g., he English a icle sys em o he simple pas ense) (Bi chene , 2008), some
esea che s ha e ound aul wi h he me i s o e o co ec ion (EC) a guing ha i s
implemen a ion may (i) lack cla i y, p ecision, and consis ency (Ellis, 2009); (ii) lack
sensi i i y on he pa o eache s o s uden s’ needs and abili y le els (Hyland, 1998); (iii)
p oduce con usion, anxie y and s ess among lea ne s due o he o e whelming numbe o
co ec ions, making i challenging o disce n non- a ge -like elemen s (Hyland, 1998); (i )
esul in passi e ecep ion and supe icial hypo heses es ing (Adams, 2003) and ( ) lead o
supe icial and ansien memo iza ion o he explici ules o co ec ions (T usco , 1996).
Consequen ly, he alue o adi ional EC o L2 de elopmen is s ill a exed ques ion
(Ma ínez Es eban & Roca de La ios, 2010). O he eedback echniques should be deployed a
leas as pa ial op ions o adi ional eedback me hods. One o he al e na i es ha has d awn
special a en ion among esea che s a e model ex s, since he e is e idence ha hey play an
ins umen al ole in p omo ing no icing and me alinguis ic awa eness, and in engaging s uden s
in deepe p ocessing han mo e adi ional eedback me hods such as EC (Hanaoka, 2007;
Hanaoka & Izumi, 2012; Ma ínez Es eban & Roca de La ios, 2010).
Models a e a ype o eedback which consis s o p o iding lea ne s wi h na i e o na i e-like
ex s ha hey compa e wi h hei o iginal d a (Ma ínez Es eban & Roca de La ios, 2010).
This mo e discu si e eedback me hod does no explici ly single ou e o s bu ea s he ex
as a whole p o iding app op ia e language, o ganiza ion, mechanics, s yle and ideas o a gi en
con ex , a he han o e ing lis s o co ec ed e o s, edi ing symbols o me alinguis ic codes
(Cáno as Gui ao, 2017).Models also appea o s imula e cogni i e con lic by p esen ing
in o ma ion ha con adic s s uden s' hypo heses abou language mechanics, p omp ing hem
o con empla e al e na i e o ms o exp essing hei in ended meanings (Coyle & Roca de
7
La ios, 2020). Because e o s a e no explici ly indica ed, modeling has also p o en e ec i e
in encou aging lea ne s o iden i y hei own mis akes, po en ially leading o mo e in-dep h
cogni i e p ocessing (Sachs & Polio, 2007). In ac , he p ima y ad an age o models lies in
c ea ing an op imal en i onmen o s uden s o ecognize he likenesses and dispa i ies
be ween hei in e language and he a ge language, p omp ing a e-e alua ion and a i ma ion
o hei L2 comp ehension and consequen ly, a e inemen o hei ou pu (Sachs & Polio,
2007). Fu he mo e, when in e ac ing wi h he na i e-like model ex , lea ne s a e no only
capable o ec i ying hei limi a ions bu also ha e he oppo uni y o encoun e na i e-like
ocabula y, g amma , idea ional, and o ganiza ional s uc u es. This exposu e se es as an
impe us o inco po a ing such elemen s in o hei subsequen ex s, aiming o con ey meaning
mo e e ec i ely. Consequen ly, i con ibu es o hei w i ing p o iciency, ex ending beyond
me e e o co ec ion. Fo example, Coyle and Cáno as Gui ao (2019) highligh ed he use o
longe sen ences on he pa o he s uden s, while Villa eal and Láza o-Iba ola (2022)
obse ed mo e syn ac ically complex s uc u es o ap ly encode he in ended message.
Wha is mo e, al hough Ellis (2016) no es ha explici WCF is mo e e ec i e han implici
WCF in ha he o me is mo e likely o assu e a en ion o o m, he a gues ha implici o ms
o WCF (such as models) ha e also p o ed o be bene icial and may e en ha e a g ea e long-
e m e ec . On he o he hand, acco ding o Villa eal and Láza o-Iba ola (2022, p. 2), model
ex s can be seen as b idging " he bounda y be ween di ec and indi ec o ms,”. So, we could
a gue ha hey migh possess he ad an ages o bo h app oaches.
Taken oge he , models would seem o se e as a aluable eedback ool o EFL lea ne s.
No wi hs anding his g ea pedagogical appeal, he e is a small body o li e a u e ha is
conce ned wi h he impac o models on he child en’s d a s.
2.3. Empi ical esea ch on models
8
The ypical sequence o esea ch wi h models consis s o engaging lea ne s in a h ee-s age
w i ing ask in ol ing (i) no icing o linguis ic p oblems while w i ing a pic u e-based s o y
(S age 1 (S1)), (ii) compa ison o hei ini ial d a s wi h he na i e-speake model, which may
o e solu ions o hei p oblems, and no icing (o no ) o hose solu ions (S age 2 (S2)) (du ing
his s age, lea ne s also ake no e o he di e ences be ween hei d a s and he model's
solu ions), and (iii) ew i ing o hei o iginal ex , which wo ks as an immedia e pos - es o
see whe he lea ne s inco po a e he solu ions in o he e ised ex s (S age 3 (S3)). Some
s udies ha e included a delayed pos - es which was ca ied ou a week (Luquin & Ga cía
Mayo, 2021) o wo mon hs la e (Hanaoka, 2006a, 2007).
So a models ha e been examined in bo h indi idual and collabo a i e w i ing wi h EFL high-
school s uden s o Spanish (Ga cía Mayo & Loidi Labandiba , 2017; Ma ínez Es eban & Roca
de La ios, 2010; Mon ealeg e Ramón, 2019), and Ko ean (Kang, 2020); wi h p ima y school
child en (Luquin & Ga cía Mayo, 2020, 2021; Cáno as Gui ao e al., 2015; Coyle & Roca de
La ios, 2020; Coyle e al., 2018; Láza o-Iba ola, 2021; C iado e al., 2022; Láza o-Iba ola &
Villa eal, 2021; Roo hoo e al., 2022; Villa eal & Láza o-Iba ola, 2022); and wi h Japanese
ESL (Abe, 2008) and EFL (Hanaoka, 2006a, 2007) uni e si y s uden s. Models ha e also been
compa ed o e o mula ions wi h wo Japanese EFL lea ne s (Hanaoka, 2006b), and wi h
Japanese (Hanaoka & Izumi, 2012) and Chinese EFL college s uden s (Yang & Zhang, 2010).
Likewise, he use o models has been con as ed wi h EC wi h p ima y school EFL lea ne s in
Spain (Coyle & Roca de La ios, 2014).
Ou o all he s udies on models ca ied ou wi h child en, o ou knowledge, only a ew ha e
been conduc ed on how models a ec he child en’s inal d a s. The i s a emp was made
by Coyle and Roca de La ios (2014), who s udied he e ec o EC and models on he no icing
and e isions o 46 Spanish EFL child en (aged 11-12). Pa icipan s we e di ided in o ou
9
g oups and ecei ed di e en ea men s o EC and models. The esul s showed ha he g oup
ha ecei ed bo h EC and models had he highes sco es in accu acy and luency. The s udy
sugges s ha he use o EC helps child en no ice e o s, he use o models helps child en
p oduce mo e accu a e and luen language, and he combina ion o bo h can be mo e e ec i e
han ei he s a egy alone.
La e , Coyle e al. (2018) conduc ed a s udy based on Cáno as Gui ao (2017), which examined
he ole o models in collabo a i e w i ing and eedback cycles o e a i e-mon h pe iod
including a pe iod o ins uc ion. The pa icipan s we e 16 Spanish EFL child en (aged 10-11)
di ided in o a g oup ha ecei ed ins uc ion o e a pe iod o six weeks and a g oup ha did
no . The s udy ocused on he p ocess o w i ing by iden i ying he ajec o ies he child en
ollowed ac oss asks and added a long- e m dimension o p e ious one-sho s udies. The 14
ajec o ies we e ca ego ized based on he na u e o challenges (o he absence he eo )
encoun e ed by he lea ne s du ing he c ea ion o hei ini ial d a s. These encompass a single
ajec o y whe e no ela ed ou pu was ini ially p esen , h ee ajec o y g oups illus a ing he
de elopmen o lea ne s' un esol ed, un epo ed, and esol ed p oblems (, and a concluding
ajec o y add essing he inco po a ion o new con en du ing ex e ision. Each ajec o y
ook a dis inc cou se, shaped by he p ocessing mechanisms ini ia ed by he lea ne s a a ious
s ages o he ask. The au ho s ound ha he gains in accep abili y and comp ehensibili y o
he child en's w i en ou pu we e achie ed as a esul o using ajec o ies wi h g ea e po en ial
o p omo e language lea ning. The s udy ce ainly opened a new dimension owa d he
p ocessing o WCF bu he small numbe o pa icipan s and he lack o s a is ical analysis
p e en s he ex apola ion o he esul s. Also, imp o emen on he pa o he eaching g oup
could also be a ibu ed o a ask- epe i ion e ec and no only o p olonged aining.
16
usage o e b enses, and mo phological aspec s such as subjec - e b ag eemen . Addi ionally,
hey add essed conce ns ela ed o he cohe ence, cohesion, and s ylis ic elemen s o he ex .
In he hi d s age, he ew i ing s age, e e y pai ecei ed he same pic u e p omp again (no
model ex o ini ial d a s we e p o ided his ime). The child en we e hen asked wi h
ew i ing he s o y, aiming o ecollec and in eg a e he aspec s hey had iden i ied in he
eedback om he p io week. The s uden s we e unawa e o his ask in ad ance, as his
app oach aimed o p e en he memo iza ion o co ec ions. The inal s age, o delayed pos -
es , in ol ed he p oduc ion o a ex based on a new isual p omp (see Appendix C). This
p omp na a ed a s o y simila o he i s one and con ained he same numbe o ca oons. The
aim was o allow s uden s o inco po a e some o he ea u es p esen in he ini ial s o y. We
delibe a ely chose a di e en isual s imulus ha was no oo dissimila o ensu e ha he asks
emained isomo phic. The pu pose behind his app oach was wo old. Fi s ly, we wan ed o
assess how much o he eedback he s uden s could emembe and e ain in he sho e m.
Secondly, we aimed o dis inguish po en ial ask- epe i ion e ec s om ac ual lea ning om
he model. This ensu ed ha any new linguis ic knowledge he child en migh ha e acqui ed
and ecalled was no solely due o pe o ming he same ask wice. Be ween Feb ua y and May,
bo h he LTG and he CG ecei ed a single pic u e p omp e e y mon h and unde ook s ages
1 (w i ing), 2 (compa ison o sel -co ec ion), and 3 ( ew i ing) o each pic u e. These
in e media e sessions we e led by hei co esponding eache s.
3.4. Da a analysis
The pa icipan s’ collabo a i e ex s (a o al o 180 ex s: 30 om S1, 30 om S3, and 30 om
S4 in Cycle 1, and 30 om S1, 30 om S3, and 30 om S4 in Cycle 2) p oduced h oughou
he s ages in bo h cycles we e ansc ibed and analyzed conside ing he ollowing aspec s: Type
o clause, CAF and holis ic measu es. The inco po a ions o eedback in o e ised ex s a e
17
ypically coded as ei he accu a e o no , as documen ed in p e ious s udies wi h adul s (see,
o example, Yang & Zhang, 2010). This me hodological decision comple ely misses ou all
he pa ial gains ha child en may ob ain om WCF, e en mo e so when he ways in which
YLs espond o eedback a e much less egula and s able. Al hough we decided o eplica e
o he s udies (e.g., Cáno as Gui ao, 2017; Coyle & Roca de La ios, 2014; To as, 2005) in he
analysis o he child en’s w i en ou pu , we ound i impo an o use much mo e nuanced
app oaches and a a ie y o hem o cap u e mino gains ac oss ex s (Coyle & Roca de La ios,
2014; Ga cía He nández e al., 2017; Villa eal & Láza o-Iba ola, 2022). The ex s, p oduced
by lea ne s wi h limi ed and insu icien knowledge ha in many cases limi s he abili y o o m
p oposi ions and a icula e hem consis en ly, o ced he adop ion o analy ical c i e ia om he
idiosync a ic pe spec i e o he na u e o he ex s hemsel es.
To assess he immedia e e ec o he models on he child en's w i ing, he quali y o he hi d
d a (S age 4) was compa ed o he ini ial d a (S age 1) in bo h cycles (compa ing d a 1 o
d a 3 and d a 4 o d a 6). We op ed o jux apose he ini ial d a wi h he delayed pos - es
d a , eschewing he in e media e S age 3 d a o mi iga e po en ial ask- epe i ion e ec s. Ou
objec i e was o sc u inize he e olu ion o s uden s as w i e s, ocusing on he manne and
ace s h ough which hey e ined hei skills, a he han assessing hei compe ence as me e
eplica o s o he iden ical ask. To s udy he long- e m impac o eedback, he beginning and
inal d a s (compa ing d a 1 o d a 6) we e compa ed. A mixed ANOVA was employed o
he analysis, inco po a ing bo h a be ween-g oups a iable (g oup) and a wi hin-g oups
a iable ( ime). To pinpoin di e ences, pos -hoc es s using he Bon e oni co ec ion we e
applied o adjus he alpha le el o he numbe o compa isons, minimizing he isk o Type 1
e o . Fo co ela ed a iables, he Mauchly es was conduc ed o e i y he assump ion o
sphe ici y. In cases whe e his assump ion was no me , he G eenhouse-Geisse es was
18
employed o ec i y he lack o sphe ici y. The signi icance h eshold was es ablished a α =
0.05.
E ec sizes we e also compu ed o each s a is ical p ocedu e. Pa ial e a squa ed was used o
gauge he e ec size o bo h one-way and mixed ANOVA. Cohen's (1988) ecommenda ions
we e ollowed, ca ego izing e ec sizes be ween 0.06 and 0.1 as 'small,' 0.15 as 'medium,' and
be ween 0.15 and 1 as 'la ge.' Pos -hoc es s we e e alua ed using he d amily o e ec sizes.
As ou lined by Cohen (1992), an e ec size o d = 0.2 co esponds o a 'small' e ec , 0.5
signi ies a 'medium' e ec , and 0.8 indica es a 'la ge' e ec .
3.4.1. Type o clause
Building upon he wo k o To as (2005) and Cáno as Gui ao (2017), we segmen ed he
pa icipan s' ex s in o clausal uni s based on hei g amma ical co ec ness. This coding
scheme was chosen o i s sui abili y in conduc ing a comp ehensi e analysis o w i en ou pu
om young EFL lea ne s, cha ac e ized by simplici y and b e i y in hei ex s. Th ee uni s
we e iden i ied: p e-clause, p o o-clause and clause. They a e de ined and exempli ied as
ollows:
P e-clause: g amma ically inco ec uni o language consis ing o agmen ed o
dis o ed s ings o wo ds, a imes incomple e, in which he meaning in en ion is no
always appa en .
Example: And ough wi h he plum
19
P o o-clause: Linguis ic uni in which he child en’s meaning in en ion is clea bu
which con ains g amma ical inaccu acies o gaps in he clausal uni .
Example: Then go o he amily o Helen
Clause: g amma ically accu a e uni o language which may p esen a sligh inaccu acy
in spelling, lexis, g amma o conco dance.
Example: One day mo ning a six o’clock, Lucy was sliping
In o de o ack he child en’s w i ing de elopmen , he o al numbe o uni s o each clause
ype was allied and compa ed in hei i s and e ised d a s in bo h cycles and ac oss g oups.
3.4.2. CAF measu es and lexical di e si y
Wi h he pu pose o iden i ying po en ial p og ess in he linguis ic accep abili y and
comp ehensibili y o he lea ne s’ w i en ex s om hei o iginal o hei e ised ex s in bo h
cycles, we ollowed To as (2005), To as e al. (2006), Cáno as Gui ao (2017) and Coyle and
Roca de La ios (2014), since hey also analyzed he w i en ou pu o child EFL lea ne s. The
measu es we e classi ied in o ou main a eas: (i) accu acy, (ii) luency, (iii) syn ac ic
complexi y and (i ) lexical di e si y.
(i) Accu acy: Following Cáno as Gui ao’s (2017), an e o a io was used o measu e
o e all accu acy: [numbe o linguis ic e o s/ o al numbe o wo ds] × 10. Like
Cáno as Gui ao (2017), we decided o use a 10-wo d a io a he han he 100-wo d
a io as he child en’s ex s we e ela i ely sho (i.e., less han 100 wo ds). E o
a ios we e compu ed and compa ed as displayed in Table 1 below.
20
Table 1
Iden i ica ion and e ision o e o s
ERROR RATIOS
O iginal ex Re ised ex
Today i is Monday and he sun is going up
Ana is sliping a 6.00 a.m. he clock has
s a ed o ing. Ana don’ wan o woke up
bu she has o go o he shool bu he clock
do hings o Ana wake up. Now, Ana is
b ushing he head and washing he ee s.
Suden ly, she akes he shool bag and goes
o he shool.
Pai 29, DTG, S age 1, Cycle 1
Today is Monday mo ning and Ma ine is
sleeping in he bed a six o´clock he clock´s
ala m s a s inging bu Ma ine isn´ wan o
wake up. Ma ine pu s he ee on he pillow,
bu is sleep. Now, he clock akes a ea he
and s a s ouching he ee o she ge up o
go o school. Now Ma ine´s washing he
ee h and combing he hai . She akes he
school bag and goes o school.
Pai
29, DTG, S a
g
e 3, C
y
cle 1
Nº wo ds: 65
Nº e o s: 20
E o a io: (20/65)x10=3,08
Nº wo ds: 66
Nº e o s: 8
E o a io: (8/66)x10=1,21
Today (1) i is Monday and he sun is going
up (2). Ana is (3) slieeping (4) and a 6.00
a.m. he clock (5) has s a ed s a s o ing.
Ana (6) don’ doesn’ wan o (7) woake up
bu she has o go o (8) he (9) shool school
(10). (11) bu he clock (12) does hings o
(13) Ana wake up Ana. Now, Ana is (14)
b ushing combing he (15) head hai and
washing b ushing he (16) ee s ee h. (17)
Suden ly suddenly, she akes he (18)
shool school bag and goes o (19) he (20)
shool school.
Today is Monday mo ning and Ma ine is
sleeping in he bed (1). a six o´clock he (2)
clock´s ala m clock s a s inging bu
Ma ine (3) isn´ doesn’ wan o wake up.
Ma ine pu s he (4) ee oo on he pillow,
bu (5) is sleep con inues sleeping. Now, he
clock akes a ea he and s a s ouching he
(6) ee oo (7) o she ge up o go o school
o wake he up (because she has o go o
school). Now Ma ine´s (8) washing
b ushing he ee h and combing he hai . She
akes he
school
b
a
g
an
d
g
oes o school.
(ii) Fluency: he o al numbe o wo ds pe ex was also conside ed. An online ex
analysis ool (h ps:// ex inspec o .com/wo k low) was used o wo d calcula ion.
(iii) Syn ac ic complexi y: Following To as e al. (2006), syn ac ic complexi y was
measu ed as numbe o subo dina e and coo dina e clauses o e o al numbe o
clauses. Table 2 ea u es examples o how luency and syn ac ic complexi y we e
coded.
21
Table 2
Example o luency and syn ac ic complexi y codi ica ion
One day a scien is was doing a po ion while is dog was sleeping. When he inishes his po ion, he
was exci ed o as e i and he d inks he po ion. When he d inks he po ion he s a ed eeling bad.
Suddenly a b igh ligh apie and a loud sound sounds and he u n in o a ca . When he dog hea d
he sound i woke up and s a ed igh ing wi h he ca . ‘Miau’ we e he las wo ds o he scien is
ca .
Pai
13, TG, S age 3, Cycle 2
Subo dina e clauses:
- While his dog was sleeping
- When he inishes his po ion
- When he d inks he po ion
- When he dog hea d he sound
Coo dina e clauses:
- and he d inks he po ion.
- and a loud sound sounds
- and he u n in o a ca .
- and s a ed igh ing wi h he ca .
To al wo d coun : 78
Numbe o sen ences1: 6
Numbe o clauses: 14
Numbe o subo dina e clauses: 4
Numbe o coo dina e clauses: 4
Clauses pe sen ence: [14/6]=2,33
(i ) Lexical di e si y: some ex ensi ely used measu es such as lexical complexi y
(numbe o e bs, adjec i es, noun, e c. ypes) as used by To as e al. (2006) o
lexical densi y (wha p opo ion o he ex con ains lexical wo ds) we e disca ded.
This me hodological decision was mo i a ed by he ac ha he ocus o he p esen
analysis was no o ob ain he o al numbe o lexical wo ds o he di e en ypes
o lexical ca ego ies p esen in he child en’s ex s, bu a he o explo e how many
di e en wo ds appea in each ex . Consequen ly, lexical di e si y (o lexical
ichness) was used as a measu emen o newly lea ned ocabula y, since we
conside ed ha i migh be e e lec he de elopmen o he child en’s
in e language.
1 A maximal clause, which con eys an independen meaning
22
Lexical di e si y is usually calcula ed using a ype- oken a io (TTR). A high TTR indica es a
high deg ee o lexical a ia ion. Al hough he TTR can be an ex emely use ul measu emen
o calcula ing he lexical di e si y o a ex , a common p oblem wi h his measu e is ha i
does no wo k as e ec i ely when dealing wi h ex s o di e en leng h. Gui aud (1960)
p oposed a measu e called ‘Roo Type Token Ra io’ (RTTR) which is ob ained by di iding he
numbe o ypes by he squa e oo o he numbe o okens hus pa ially add essing he
p oblem o TTR’s a iance on ex leng h. Each ansc ibed ex was uploaded o a so wa e
ool (h ps:// ex inspec o .com/wo k low) which would au oma ically calcula e he numbe o
ypes and okens. P io o uploading he child en’s w i en ex s, spelling mis akes we e
co ec ed as he aim o lexical di e si y is o analyze how di e se he ange o wo ds used is
and he e o e he so wa e mus ecognize he wo ds. Once his in o ma ion was p o ided, we
applied Gui aud’s (1960) o mula o he da a ( ypes/√ okens), hus ob aining he RTTR o
each ex .
3.4.3. Holis ic measu es
We also assessed he ex s bo h quan i a i ely and quali a i ely aking in o accoun measu es
o adequacy, cohe ence, cohesion, g amma ical accu acy, lexical ange and mechanics. A
h ee-poin sco ing ub ic was used o e alua e he w i ings, 3 being good, 2 a e age and 1
poo . We op ed o he ub ic de eloped by Villa eal and Muna iz-Iba ola’s (2021) (see also
Hedgcock & Le kowi z, 1992). This ub ic was chosen due o i s abili y o e ec i ely a ge
p ecise lea ning objec i es and skills, acili a ing a ho ough assessmen o he desi ed
ou comes. To ensu e a ing eliabili y, he pa icipan s’ w i en p oduc ion was coded by one
esea che and 20% o he da a was independen ly coded by a second esea che . In e - a e
ag eemen esul ed in 97%, and di e ences we e discussed un il o al ag eemen was eached.
23
4. Resul s
In o de o examine he impac o he models on he child en’s w i en p oduc ion in he sho -
e m, he quali y o he hi d d a (S age 4) in ela ion o he i s one (S age 1) was measu ed
in bo h cycles (d a 1 s. d a 3 and d a 4 s. d a 6). To examine he e ec o he eedback
in he long un, he i s and las d a (d a 1 s. d a 6) we e compa ed (see Appendix D o
a able con aining desc ip i e s a is ics, including he ype o clause and CAF measu es,
o ganized by g oup and d a ).
As o p e-clauses, wi hin he global domain o ype o clause, he G eenhouse-Geisse
co ec ion was used since he sphe ici y assump ion was no me (X2 = 13.74; p = .017). Resul s
e ealed a main e ec o P e-clause (F(2.21,59.72) = 16.92; p = < .000; ηp2 = 0.38), G oup (F(2,27)
= 3.74; p = .037; ηp2 = 0.21), and o he P e-clause x G oup (F(4.42,59.72) = 8.03; p = .048; ηp2 =
0.07) in e ac ion. Pai wise compa isons wi h Bon e oni adjus men showed some s a is ically
signi ican in a- and in e -subjec di e ences. As o he o me , he TG signi ican ly educed
he numbe o p e-clauses om d a s 1 o 3 (p = .004, 95% CI [0.33, 2.21], d = 0.70), 4 o 6
(p = .045, 95% CI [-0.40, 1.13], d = 0.25) and 1 o 6 (p = .024, 95% CI [0.13, 2.60], d = 0.58).
The LTG also imp o ed signi ican ly om hei i s o hei hi d composi ion (p = .004, 95%
CI [0.31, 2.29], d = 0.68), and om he i s o he las one (p = .031, 95% CI [-0.25, 3.20], d
= 0.48). The sel -co ec ion g oup showed no imp o emen in e ms o p e-clauses. G oup
di e ences we e only obse ed on he i s delayed pos - es (d a 3) whe e bo h ea men
g oups ou pe o med he CG (p = .007, 95% CI [0.28, 2.02], d = 0.61 o CG-TG; p = .002,
95% CI [0.44, 2.23], d = 0.70 o CG-LTG). In conclusion, bo h ea men g oups exhibi ed a
dec ease in he numbe o p e-clauses om d a s 1 o 3 and om d a s 1 o 6. Fu he mo e,
hey su passed he CG in he pos - es o Cycle 1 (d a 3) by p oducing a lowe coun o p e-
clauses.
24
Rega ding p o o-clauses, Mauchly’s es did no indica e any iola ion o sphe ici y (X2 = 4.07;
p = .539). A main e ec was ound o P o o-clause (F(3,81) = 4.73; p = .004; ηp2 = 0.14) and o
he in e ac ion be ween P o o-clause and G oup (F(6,81) = 2.82; p = .044; ηp2 = 0.14). Pos -hoc
analyses once mo e un eiled signi ican dispa i ies be ween d a 1 and d a 6 wi hin bo h
model g oups (p = .032, 95% CI [-1.61, 3.52], d = 0.37 o TG; p = .006, 95% CI [0.63, 4.97],
d = 0.67 o LTG), bu no o he CG, indica ing ha in he TG he e was a no able educ ion
in p o o-clauses om d a 1 o d a 6, while a signi ican dec ease was obse ed o he LTG
du ing his ansi ion. This ime, no be ween-g oup di e ences we e obse ed o his aspec .
Conce ning clauses, sphe ici y was me (X2 = 8.34; p = .139), as indica ed by Mauchly’s es .
The e is a signi ican main e ec o Clause (F(3,81) = 14.18; p = < .000; ηp2 = 0.34), o G oup
(F(2,27) = 5.52; p = .043; ηp2 = 0.12), and o he Clause x G oup (F(6,81) = 7.96; p = .027; ηp2 =
0.46) in e ac ion. Fu he pai wise compa isons e ealed s a is ically signi ican wi hin-g oup
and be ween-g oup di e ences. In his case, we ound ha he CG w o e s a is ically ewe
clauses in d a 6 in con as o d a 1 (p = .002, 95% CI [-0.48, 5.30], d = 0.41), and he LTG
p oduced a signi ican ly highe numbe o clauses in d a 6 in compa ison wi h d a 4 (p =
.012, 95% CI [-3.66, -0.34], d = -0.63). No di e ences ac oss d a s we e obse ed o he TG.
Looking a be ween-g oup di e ences, pos -hoc analyses e ealed ha he LTG ou pe o med
bo h he CG (p = .001, 95% CI [1.30, 5.21], d = 0.78) and he TG (p = .014, 95% CI [0.39,
4.11], d = 0.56) in d a 6. In summa y, we no ed a dec ease in clause p oduc ion om he
ini ial o he inal e sion o he ex s wi hin he CG. Con e sely, o he LTG g oup, a
signi ican inc ease in he numbe o clauses was obse ed om d a s 4 o 6. Ac oss all g oups,
he LTG g oup gene a ed a signi ican ly highe numbe o clauses in d a 6 compa ed o he
o he wo eedback condi ions.
25
Mo ing on o he a ea o complexi y, and mo e speci ically o subo dina e clauses, we used he
G eenhouse-Geisse co ec ion since he sphe ici y assump ion was ejec ed (X2 = 20.13; p =
.001). Mixed ANOVA showed a main e ec o Subo dina e clause (F(2.07,56.10) = 11.49; p = <
.000; ηp2 = 0.29), and o he Subo dina e clause by G oup (F(4.15,56.10) = 9.62; p = .037; ηp2 =
0.15) in e ac ion e ec . Fu he pos -hoc es s loca ed hese di e ences o bo h expe imen al
g oups be ween d a s 1 and 3 (p = .006, 95% CI [0.45, 3.55], d = -0.67 o TG; p = .024, 95%
CI [-2.29, 1.09], d = -0.18 o LTG) and 4 and 6 (p = .048, 95% CI [0.62, 3.16], d = 0.37 o
TG; p = .001, 95% CI [1.34, 3.22], d = 0.13 o LTG), which means ha ollowing he eceip
o he models in bo h cycles, speci ically on bo h pos - es s, he ea men g oups in eg a ed a
g ea e numbe o subo dina e clauses when compa ed o hei ini ial d a s in S age 1. No
s a is ically signi ican di e ences we e ound o he CG o be ween g oups.
Wi h e e ence o coo dina e clauses, he assump ion o sphe ici y was no ejec ed (X2 = 10.93;
p = .053). The esul s ob ained om he analysis o coo dina e clauses only e ealed a
signi ican main e ec o Coo dina e clause (F(3,81) = 18.11; p = < .000; ηp2 = 0.40).
In e ms o lexical di e si y, Mauchly’s es s indica ed ha sphe ici y was no iola ed (X2 =
6.27; p = .281). We obse ed a signi ican main e ec o Lexical di e si y (F(3,81) = 23.18; p
= < .000; ηp2 = 0.46), G oup (F(2,27) = 9.33; p = .028; ηp2 = 0.09), and o he in e ac ion be ween
Lexical di e si y x G oup (F(6,81) = 16.43; p = .021; ηp2 = 0.17). Fo he Bon e oni pos -hoc
es , he e is a s a is ical di e ence be ween d a s 1 and 6 wi hin he h ee g oups. Tha is, he
lexical epe oi e o he child en in bo h he CG (p = .013, 95% CI [10.53, 123.03], d = 0.62)
and he TG (p = .002, 95% CI [23.21, 124.97], d = 0.76) appea ed o be signi ican ly iche in
d a 1 as opposed o d a 6. On he con a y, he LTG (p = .004, 95% CI [18.84, 125.56], d =
0.70) appea ed o inc ease he quan i y o lexical wo ds, and his ise was ound o be
32
inc ease in clauses wi hin such a sho ime ame may suppo he idea ha he g ea e he
exposu e o models, he g ea e he bene i s he child en may ob ain, a leas in ce ain linguis ic
aspec s. Al hough u he s udy is wa an ed gi en he scan esea ch exis ing o d aw a i m
conclusion, hese posi i e indings hin a a possible ela ion be ween models and he linguis ic
accep abili y o he child en’s w i en ou pu .
Exposu e o he model ex wi hin he i s and second w i ing cycles was also ound o ha e
posi i e sho - e m e ec s on enhancing he syn ac ic complexi y o he child en’s hi d d a ,
which is isible in he highe numbe o subo dina e clauses p oduced by he ea men g oups
on bo h pos - es s. No di e ences ac oss g oups o cycles we e ound. No di e ences we e
obse ed ei he when analyzing he e olu ion o lexical di e si y indica o s om p e- o pos -
es s in each g oup, bu he esul s e ealed ha he LTG showed highe lexical di e si y han
he o he wo g oups bo h when hey ini ia ed (d a 4) and inished (d a 6) Cycle 2. The ac
ha hey ou pe o med he o he wo g oups as soon as Cycle 2 began mani es s once again a
posi i e ela ionship be ween epea ed exposu e o models and he a ie y o wo ds used in he
child en’s ex s. As a ma e o ac , long- e m esul s e ealed ha he numbe o di e en
wo ds used by he lea ne pai s in he CG and he TG dec eased signi ican ly om d a 1 o
d a 6. On he con a y, he g oup bene i ing om models du ing ou mon hs showed a
signi ican imp o emen , isible om hei i s o hei las w i en ex . I we conside o he
simila s udies, we obse e mixed esul s. Fo example, in hei c oss-sec ional s udy wi h
child en, Villa eal and Láza o-Iba ola (2022) obse ed ha he model g oup w o e mo e
g amma ically and lexically complex ex s on he pos - es . This inding, howe e , does no
suppo ha o Láza o-Iba ola (2021), who did no epo any ex imp o emen s in e ms o
CAF wi h a simila sample and a e ha ing e ised he exac same models as in Villa eal and
Láza o-Iba ola’s (2022) s udy. Fo he pa , Cáno as Gui ao (2017) e ealed ha all child en
33
w o e less g amma ical ex s han in he i s w i ing cycle, whe eas Roo hoo e al. (2022)
obse ed ha he child en’s d a s signi ican ly imp o ed in lexical di e si y, bu no in
syn ac ic complexi y, wi h models. Gi en he con lic ing esul s, mo e ex ual analyses a e
needed o de e mine whe he he bene i s associa ed wi h models lead o gains in lexical
di e si y and syn ac ic complexi y.
Wi h espec o accu acy, no meaning ul sho - e m indings we e obse ed. The only
s a is ically signi ican esul was o he LTG, who made signi ican ly ewe mis akes om
d a 1 o d a 6, and di e ences we e also obse ed wi h espec o he CG and he TG in bo h
d a s 4 and 6. The sho - e m indings co ela e a o ably wi h Villa eal and Láza o-Iba ola
(2022) and Láza o-Iba ola (2021), who did no ob ain any ema kable gains o any o he
g oups. The long- e m esul s, howe e , somewha mi o he indings ob ained in he measu es
o ype o clause, whe e we obse ed a signi ican ly highe numbe o clauses o accep able
uni s a e a long ea men wi h models. This lack o impac on accu acy in he sho un may
suppo he hypo heses o he child en’s de eloping me alinguis ic abili y and he p imacy o
meaning o e o m (VanPa en, 2004). Fu he mo e, model ex s ha e ga ne ed p aise
p ima ily o hei emphasis on lexis (Chandle , 2003; Coyle & Roca de La ios, 2014), while
also being linked o syn ac ic complexi y ad an ages as no ed by Villa eal and Láza o-Iba ola
(2022). Wha e e he eason, one mo e ime hese obs acles we e o e come by he child en
who had access o se e al models o he pas ou mon hs, as hey managed o p oduce mo e
accu a e and hus mo e accep able and comp ehensible ex s.
In line wi h p e ious esea ch (Cáno as Gui ao, 2017; Coyle & Roca de La ios, 2014; Láza o-
Iba ola, 2021; Roo hoo e al., 2022; Villa eal & Láza o-Iba ola, 2022), luency appea s o
be esilien o models. The simila i y in he o al numbe o wo ds w i en be o e and a e
34
exposu e o he wo eedback echniques indica es ha nei he o m o eedback has much
impac , a leas in he sho e m, on he leng h o he ex s. I seems ha he need o con ey
meaning os e s quali a i e changes, bu no quan i a i e ones (Coyle & Roca de La ios, 2014;
Láza o-Iba ola, 2021; Villa eal & Láza o-Iba ola, 2022). Coyle and Roca de La ios (2014)
asc ibe his s abili y o luency ac oss d a s o he b ie ness o he ex s he child en we e asked
o w i e. In a simila ein, Villa eal and Láza o-Iba ola (2022) a ibu e his ou come o ask
ype, which clea ly di ec s he child en’s a en ion o con en and inhibi s mo e c ea i e
a emp s. Howe e , his inding does no coincide wi h C iado e al. (2022), whose pa icipan s
in he model g oup demons a ed ad ancemen in luency and he sel -co ec ion g oup
exhibi ed e en g ea e luency han he eedback g oup. I is possible ha he p e ious pen-
and-pape s udies a e a a disad an age when compa ed o he digi al w i ing used by C iado
e al. (2022). This p omp s specula ion ha digi al w i ing migh hold an uppe hand o e
adi ional pape -based w i ing, as sugges ed by Villa eal e al. (2021).
No wi hs anding he abo e, we did ind long- e m changes in he numbe o wo ds p oduced.
When compa ing he i s and las ex s, luency unde wen a p onounced decline in he h ee
g oups, ega dless o he eedback condi ion. We in e p e his sha p d op in he child en’s
ex ual luency in ela ion o he end o academic yea . The pa icipan s we e clea ly i ed and
eage o inish he asks. The majo i y o pa icipan s exhibi ed a gene al lack o concen a ion
and mo i a ion, as p e iously obse ed in he s udy by Kopinska and Azka ai (2020), esul ing
in b ie dialogues and sho e ex s, which was also epo ed by Cáno as Gui ao (2017).
Finally, as o he global analysis o d a quali y, he holis ic sco es showed an ac oss-cycle
imp o emen o he TG’s ex ual cohesion whe eas, in line wi h he quan i a i e analysis, he
LTG enhanced hei lexical epe oi e om d a 1 o d a 6. When he g oups we e con as ed,
35
he TG seemed o do signi ican ly be e han he CG in e ms o cohe ence on he i s pos -
es , while he LTG ob ained a signi ican ly highe sco e in cohesion han he CG and he TG
on he second pos - es . Láza o-Iba ola (2021) and Roo hoo e al. (2022) also ound a no able
imp o emen in he o e all e alua ion o p ima y school s uden s who examined hei w i en
pieces agains model ex s, while Villa eal and Láza o-Iba ola (2022) did no obse e any
signi ican di e ences. The combina ion o di e en measu emen s when analyzing w i en
p oduc ion is key o be able o g asp any imp o emen s in d a quali y. Howe e , we belie e
ha sho - e m upg ading may no be as de ec able as long- e m upg ading when i comes o
holis ic analysis, since any mino gain ob ained in wo weeks’ ime may no be humanly
no iceable. In addi ion, because he a bi a iness o his ool alongside he sho ange o alues
p o ided in he ub ic used in his s udy made i di icul o gua an ee a eliable assessmen on
he child en’s w i en ex s, he esul s om such analyses should hus be ea ed wi h he
u mos cau ion and only be conside ed as an add-on analysis o he quan i a i e assessmen .
In sum, he sho and long- e m gains obse ed h ough ype o clause, CAF and holis ic
analyses o e u he empi ical e idence ha models can and do help L2 lea ne s upg ade hei
w i en ou pu and de elop hei eme ging in e language. We can hus con i m ha models help
imp o e he o e all w i en p oduc ion o p ima y EFL s uden s in bo h he sho and long un.
Mo e speci ically, in he sho - e m, models helped he child en educe he numbe o p e-
clauses and p o o-clauses as well as inc ease he syn ac ic complexi y o hei ex s. A e a
long exposu e o models, he child en we e able o (i) p oduce ewe p e-clauses, ewe p o o-
clauses and mo e clauses, (ii) use a highe di e si y o lexical wo ds in hei ex s and (iii) make
ewe e o s. As was he case wi h inco po a ions, p ac ice wi h model ex s appea ed o
acili a e he eedback p ocessing, dec easing i s complexi y and enhancing i s no icing
p ocesses. This inc eased no icing hen esul ed in upg ading o hei w i en ex s and
36
consequen ly in L2 de elopmen . None heless, we belie e ha p ac ice is no enough when i
comes o YLs. I we wan child en o make he mos ou o models, we need consciousness
aising ac i i ies o help younge and weake lea ne s enhance hei me a-awa eness o
language.
When compa ed o p e ious esea ch, ou indings conce ning he e ec o model ex s on ex
quali y among child en cas a new ligh on he con o e sy abou which aspec s o language
a e mos bene i ed bu also help o ein o ce he iew ha models do ha e a a o able e ec
on he child en’s w i en ou pu . This s udy also unde sco es he need o include a wide a ie y
o ine-g ained measu es no o miss ou any mino imp o emen and he explo a ion o
di e en ools ha co obo a e whe he small di e ences do o do no show a po en ial end.
6. Conclusion
The p esen s udy o e s he i s longi udinal assessmen o models wi h a subs an ial sample
size o pa icipan s, enabling obus s a is ical analyses o s eng hen he indings ega ding he
endu ing ad an ages o model ex s on EFL child en's w i en p oduc ion.
The esul s o his s udy highligh he need o young EFL lea ne s o engage in w i ing
ac i i ies and o ecei e and assimila e eedback on hei w i en wo k in o de o enhance hei
g asp o he L2 (Láza o-Iba ola, 2023). In EFL con ex s, language eache s o en ely on
ex books ha os ensibly p omo e communica i e language ins uc ion, o e ing a ange o
exe cises encompassing o mal aspec s, ocabula y, and eading, all o ganized hema ically.
Howe e , w i ing equen ly akes a back sea , o en ecei ing less a en ion han o al asks.
Consequen ly, i becomes impe a i e o eache s o ecognize he ad an ages o w i ing and
37
eedback p ocessing in L2 lea ning, as well as he heo e ical implica ions o w i ing- o-lea n,
as highligh ed by Cáno as Gui ao (2017) and Manchón (2009).
Model ex s ep esen jus one o m o eedback echnique, a he han he sole op ion a ailable.
Ins ead o app oaching al e na i e and adi ional WCF as mu ually exclusi e, eache s can
combine hese wo app oaches o se e dis inc pu poses. Fo ins ance, hey could o e
un ocused and/o indi ec eedback on ini ial d a s, di ec ing a en ion o a wide a ay o
aspec s and acili a ing sel -co ec ion. Subsequen ly, di ec and/o ocused eedback could be
supplied o e ised d a s, allowing child en o ec i y emaining e o s and p oduce mo e
p ecise ex s.
P io i izing an unde s anding o he ad an ages inhe en in di e se eedback me hods and
de e mining which echnique aligns bes wi h he needs o ou lea ne s should be a key
conside a ion o language ins uc o s.
Ou esea ch has some sho comings ha need o be acknowledged, al hough we belie e ha
hese limi a ions could be a sp ingboa d o u u e wo k on he opic. The s udy is limi ed i s
and o emos because o he me hodological decisions conce ning he analysis o he child en’s
w i ing. The ex s p oduced by child en a ini ial s ages o L2 lea ning o ced he esea che s
o adop analy ical c i e ia om he pe spec i e o he na u e o he ex s hemsel es. The e o e,
he low quali y and b ie ness o he w i en ex s mean ha he ins umen s and p ocedu es
no mally used in esea ch wi h olde lea ne s could no be used. Fu he mo e, he use o dis inc
measu es hampe s he possibili y o making compelling compa isons. Al hough we decided o
eplica e o he s udies o he analysis o he child en’s w i en ou pu and used he clausal uni
measu e o cap u e mino gains ac oss ex s, we also wan ed o include CAF measu es o yield
mo e de ailed in o ma ion. E en so, he p esen s udy may all sho when i comes o p o iding
a ull and accu a e pic u e o he child en’s w i ing de elopmen . Simila ly, he a bi a iness o
38
he ub ic used in he holis ic analysis alongside he sho ange o alues p o ided in i made
i di icul o gua an ee a eliable assessmen on he child en’s w i en ou pu , and he e o e he
esul s ob ained om such analyses should be ea ed wi h cau ion and only be conside ed as a
complemen a y analysis o he quan i a i e assessmen .
Finally, i is impo an o ecognize ha in his s udy models se ed as a means o an ul ima e
goal. The no iceable gains ac oss di e en con ex s a e e iden ly shown only in a speci ic
pic u e na a ion ask. Howe e , his insigh implies ha he obse ed long- e m imp o emen s
a e indica i e o he po en ial o child en as p ospec i e EFL w i e s. Sus ained exposu e o
models (and al e na i e eedback echniques) o e ime may os e his g ow h. Consequen ly,
conduc ing ollow-up asks o del e in o he ami ica ions o hese gains p esen s a p omising
di ec ion o u u e esea ch.
39
APPENDICES
Appendix A: Pic u e s o y S1
(Taken om Lapkin e al., 2002)
40
Appendix B: Model ex S2
Ma ine’s ala m clock
(Taken om Lapkin, Swain & Smi h, 2002)
41
Appendix C: Pic u e s o y S4
48
Long, M.H. (2014). Second language acquisi ion and ask-based language eaching. London:
Rou ledge.
Luquin, M., & Ga cía Mayo, M. P. (2020). Collabo a i e w i ing and eedback: An explo a o y
s udy o he po en ial o models in p ima y EFL s uden s’ w i ing pe o mance.
Language Teaching o Young Lea ne s, 2(1), 73–100.
Luquin, M., & Ga cía Mayo, M. P. (2021). Explo ing he use o models as a w i en co ec i e
eedback echnique among EFL child en. Sys em, 98(1), 1–13.
Lys e , R. (2007). Lea ning and Teaching Languages h ough Con en . Ams e dam: John
Benjamins.
Manchón, R. M. (2009). B oadening he pe spec i e o L2 w i ing schola ship: The
con ibu ion o esea ch on o eign language w i ing. In R. M. Manchón (Ed.), W i ing
in o eign language con ex s. Lea ning, eaching, and esea ch (pp. 1-19). Cle edon:
Mul ilingual Ma e s.
Manchón, R. M. (Ed.) (2011). Lea ning- o-W i e and W i ing- o-Lea n in an Addi ional
Language. Ams e dam: John Benjamins.
Ma ínez Es eban, N., & Roca de La ios, J. (2010). The use o models as a o m o w i en
eedback o seconda y school pupils o English. In e na ional Jou nal o English
S udies, 10, 143–170.
Mon ealeg e Ramón, F. (2019). Explo ing he lea ning po en ial o models wi h seconda y
school EFL lea ne s. Spanish Jou nal o Applied Linguis ics, 32(1), 155-184.
Muñoz, C. (Ed.). (2006). Age and he Ra e o Fo eign Language Lea ning. Cle edon:
Mul ilingual Ma e s.
Muñoz, C. (2017). T acing ajec o ies o young lea ne s: Ten yea s o school English lea ning.
Annual Re iew o Applied Linguis ics, 37, 168-184.
h ps://doi.o g/10.1017/S0267190517000095.
49
Nicholas, H., & Ligh bown, P.M. (2008). De ining child second language acquisi ion, de ining
oles o L2 ins uc ion. In J. Philp, R. Oli e , & A. Mackey (Eds.), Second language
acquisi ion and he younge lea ne : Child’s play? (pp. 27–52).
Ams e dam/Philadelphia: John Benjamins.
Polio, C. (2012). The ele ance o second language acquisi ion heo y o he w i en e o
co ec ion deba e. Jou nal o Second Language W i ing, 21, 375 – 389.
Reinde s, H. (2009). Lea ne up ake and acquisi ion in h ee g amma -o ien ed p oduc ion
ac i i ies. Language Teaching Resea ch, 13(2), 201–222.
Roo hoo , H., Láza o-Iba ola, A., & Bul é, B. (2022). Task epe i ion and co ec i e eedback
ia models and di ec co ec ions among young EFL w i e s: D a quali y and ask
mo i a ion. Language Teaching Resea ch. h ps://doi.o g/j h
Sachs, R., & Polio, C. (2007). Lea ne s’ uses o wo ype o w i en eedback on a L2 w i ing
e ision ask. S udies in Second Language Acquisi ion, 29(1), 67-100.
Schmid , R. (2001). A en ion. In P. Robinson (Ed.), Cogni ion and second language
ins uc ion (pp. 3–32). Camb idge: Camb idge Uni e si y P ess.
S o ch, N. (2016). Collabo a i e w i ing. In R. M. Manchón & P. K. Ma suda (Eds.), Handbook
o second and o eign language w i ing (pp. 387–406). Be lin: De G uy e .
S o ch, N. (2021). Collabo a i e w i ing: P omo ing languaging among language lea ne s. In
M. P. Ga cía Mayo (Ed.), Wo king Collabo a i ely in Second/Fo eign Language
Lea ning (pp. 13–34). Be lin: De G uy e Mou on.
Tellie , A. & Roeh -B ackin, K. (2017). Raising Child en’s Me alinguis ic Awa eness o
Enhance Class oom Second Language Lea ning. In M. Ga cía Mayo (Ed.), Lea ning
Fo eign Languages in P ima y School: Resea ch Insigh s (pp. 22-48). B is ol, Blue
Ridge Summi : Mul ilingual Ma e s. h ps://doi.o g/10.21832/9781783098118-004.
50
To as, R. (2005). P ocesos psicolingüís icos implicados en la adquisición del inglés en el
con ex o de la enseñanza p ima ia [Psycholinguis ic p ocesses in ol ed in lea ning
English in a p ima y school con ex ]. Lenguaje y Tex os, 23, 89–112.
To as, M.R., Na és, T., Celaya, M.L., & Pé ez-Vidal, C. (2006). Age and IL de elopmen in
w i ing. In C. Muñoz (Ed.), Age and he Ra e o Fo eign Language Lea ning (p. 156-
182). B is ol: Mul ilingual Ma e s.
T usco , J. (1996). The case agains g amma co ec ion in L2 w i ing classes. Language
Lea ning, 46(2), 327-369.
VanPa en, B. (2004). Inpu p ocessing in SLA. In B. VanPa en (Ed.), P ocessing ins uc ion:
Theo y, esea ch and commen a y (pp. 5–32). Mahwah, NJ: E lbaum.
Villa eal, I., Bueno-Alas uey, M. & Sáez-León, R. (2021). Compu e -based collabo a i e
w i ing wi h young lea ne s: E ec s on ex quali y. In M. Ga cía Mayo (Ed.), Wo king
Collabo a i ely in Second/Fo eign Language Lea ning (pp. 177-198). Be lin, Bos on:
De G uy e Mou on. h ps://doi.o g/10.1515/9781501511318-008
Villa eal, I., & Láza o-Iba ola, A. (2022). Models in collabo a i e w i ing among CLIL
lea ne s in P ima y School: Linguis ic ou comes and mo i a ion ma e s. Sys em,
110(4), 1-21.
Villa eal, I., & Ma ínez-Sánchez, A. (2023). Explo ing immedia e and p olonged e ec s o
collabo a i e w i ing on young lea ne s' ex s: L2 e sus FL. IRAL In e na ional e iew
o applied linguis ics in language eaching, 61. h ps://doi.o g/10.1515/i al-2022-0062.
Villa eal, I., & Muna iz-Iba ola, M. (2021). “Toge he we do be e ”: The e ec o pai and
g oup wo k on young EFL lea ne s’ w i en ex s and a i udes. In M. P. Ga cía Mayo
(Ed.), Wo king Collabo a i ely in Second/Fo eign Language Lea ning (pp. 89-116).
Be lin, Bos on: De G uy e Mou on. h ps://doi.o g/j j
51
Wiggleswo h, G., & S o ch, N. (2012). Wha ole o collabo a ion in w i ing and w i ing
eedback. Jou nal o Second Language W i ing, 21(4), 364-374.
Williams, J. (2012). The po en ial ole(s) o w i ing in second language de elopmen . Jou nal
o Second Language W i ing, 21, 321–331.
Wu, Z., Qie, J., & Wang, X. (2023) Using model ex s as a ype o eedback in EFL w i ing.
F on ie s in Psychology, 14. doi: 10.3389/ psyg.2023.1156553
Yang, L., & Zhang, L. (2010). Explo ing he ole o e o mula ions and a model ex in EFL
s uden s’ w i ing pe o mance. Language Teaching Resea ch, 14, 464–484.
D . Ma ia Luquin holds a PhD in Second Language Acquisi ion om he Uni e si y o he
Basque Coun y (UPV/EHU). He disse a ion, which was supe ised by P o D . Ma ía del
Pila Ga cía Mayo and unded wi h a p edoc o al esea ch g an (FPI) om he Minis y o
Science, Inno a ion and Uni e si ies, was awa ded he highes ma k “cum laude”, and has
been p oposed o he Ou s anding PhD disse a ion awa d. She cu en ly holds he posi iono
Assis an P o esso a he Public Uni e si y o Na a e (UPNA). He main esea ch in e es s
lie in he use o collabo a i e w i ing asks and model ex s as a eedback echnique in
p ima y school EFL se ings. Mo eo e , she also ocuses on he impac o ac o s such as
mo i a ion o a i udes, as well as on ocus on o m. In 2021, Ma ía Luquin ca ied ou a
esea ch s ay a Conco dia Uni e si y (Mon eal, Canada), unde he supe ision o D . Kim
McDonough.
h ps://o cid.o g/0000-0002-3464-8396
D . Ma ía del Pila Ga cía Mayo (h ps://laslab.o g/s a /pila ) is Full P o esso o English
Language and Linguis ics a he Uni e si y o he Basque Coun y (UPV/ EHU). She has
published widely on he L2/L3 acquisi ion o English mo phosyn ax and he s udy o
con e sa ional in e ac ion in EFL. She has been an in i ed speake o uni e si ies in Eu ope,
Asia and No h Ame ica and is an Hono a y Consul an o he Shanghai Cen e o Resea ch
in English Language Educa ion. P o . Ga cia Mayo is he di ec o o he esea ch g oup
Language and Speech and he MA p og am Language Acquisi ion in Mul ilingual
Se ings. She is he edi o o Language Teaching Resea ch. She was a membe o AILA
Execu i e Boa d-In e na ional Commi ee and he in e na ional ela ions ep esen a i e o he
Spanish Socie y o Applied Linguis ics and cu en ly belongs o he S ee ing Commi ee o
he Spanish S a e Resea ch Agency.
h ps://o cid.o g/0000-0002-1987-4889
