The role of quantity and quality of linguistic exposure on language development during childhood

The ole o quan i y and quali y o
linguis ic exposu e on language
de elopmen du ing childhood
Jose J. Pé ez-Na a o
2023
The ole o quan i y and quali y o
linguis ic exposu e on language
de elopmen du ing childhood
PhD disse a ion
Jose J. Pé ez-Na a o
Supe ised by:
D . Ma ie Lallie
D . Nicola Molina o
Basque Cen e on Cogni ion, B ain, and Language (BCBL)
Uni e si y o he Basque Coun y (UPV/EHU)
Donos ia-San Sebas ián, 2023
(cc)2023 JOSE JAVIER PEREZ NAVARRO (cc by 4.0)
Acknowledgemen s
I would no ha e been able o comple e his wo k wi hou he help o a huge
g oup o people (in e ms o quan i y and quali y). I eel uly g a e ul o ha ing
collabo a ed wi h an inc easing g oup o people along he doc o al p og am, as each
pe son came wi h a di e en se o ideas ha made he p ojec s only mo e in o med
and be e in many ways.
Fi s o all, I would like o hank my supe iso s. I ha e seen (and hope I can
lea n) an inc edible numbe o posi i e aspec s in Ma ie, my main supe iso . On he
scien i ic side, Ma ie has shown me how o be analy ical when planning an
in es iga ion, skep ical when analyzing ou da a, and mode a e as well as s aigh o he
poin when in e p e ing ou indings. I also uly app ecia e he asse i eness o
managing my expec a ions and us a ions, he hones in e es in my g ow h as a
esea che and he eye on my well-being along he yea s. I am all in all uly g a e ul
ha I was supe ised by Ma ie, and I hope I can be a simila supe iso one day. I has
also been a pleasu e o be co-supe ised by Nicola. His ideas o analyses a e e eshing
and s imula ing, and he always inds a way o look a s udies om ou side he box and
no ge s uck wi hin a pa icula idea, expe imen al pa adigm o me hodology. I was a
g ea expe ience o collabo a e wi h him, and lea n o in eg a e ideas om di e en
pe spec i es. Many hanks also o my close collabo a o s in he p ojec s ha o m his
hesis: Mikel Liza azu, o his pa ien echnical suppo and a ailabili y o each me a
lo abou EEG esea ch; Ca he ine Cla k, o he g ea job when collec ing da a and he
insigh s abou how o imp o e such collec ion; Gio gio Piazza, o his i eless wo k in
da a collec ion and his a ailabili y o in eg a e his knowledge on speech analyses wi h
my wo k on EEG; Anas asia Klimo ich-G ay, o he a ailabili y o suppo me wi h
mTRF analyses and he esh ideas on how o concep ualize ou esul s as well as how
o go beyond he expe imen and look a such esul s wi hin a big pic u e. Many hanks
also o Mikel I uskie a, o o e ing his echnical suppo and expe ise on NLP
applica ions in ou p ojec s.
I would also lo e o hank my supe iso s in he esea ch s ays in which I spen
pa o my doc o al ime. I was a pleasu e o mee and exchange ideas wi h all he lab
ma es a he CNE, specially wi h Ta suya and João. And special hanks o my co-
supe iso o one o my doc o al p ojec s, Usha, who app oached such p ojec (and
ideas o u he esea ch) wi h he mo i a ion and cu iosi y o a pe ennial s uden .
Also, o my co-supe iso a he CNE, Sheila, who was an amazing eache and made me
unde s and wi h ease he complex ideas abou speech enginee ing. Al hough he second
esea ch s ay is no pa o he p esen hesis, I would lo e o hank all he lab membe s
o Gi audLab o hei insigh s when p esen ing my p ojec s. Specially o my supe iso
in his second s ay, I saso, who augh me so much when hinking abou language and
he b ain om a biological pe spec i e, which de ini ely shaped he way I unde s ood
my hesis and s imula ed my ideas o u u e esea ch.
Coming back o he BCBL, I would like o hank he suppo o Ana, Maide , and
Eide wi h all he bu eauc acy in ol ed in p ojec s and g an s, as hey made my wo k so
much easie . Also, he help o he IT depa men , Bo ja, Jose and Gu i. Special hanks o
he lab depa men , ha ha e made my expe imen s un smoo he han I could e en
imagine. In pa icula , I would lo e o say “mila eske ” o A ai z, Manex, Lei e, Pa icia,
Mai e, Da id, La ai z, and Oihana, ha accep ed me as “one o hei own” in he long
pe iods I spen collec ing da a in he lab. I also hank all he esea che s in he BCBL
ha ha e o e ed hei suppo in hese pas yea s, and also hose wi h which I ha e
exchanged ideas o jus spen a g ea ime wi h. Specially o Alexia An zaka, Paula Ríos-
López, Amaia Ca ión, Jose Aguas i as, Candice F ancés, Sand a Gisbe , Eugenia
Na a a, Te esa Es eban, Polina Timo ee a, Shuang Geng, Pa xi Elosegi, Sa a Guediche,
Ma ina Kalashniko a, Sendy Ca a a, Ma ia Kou sougiannakis, Manuela Ruzzoli,
S e lana Pine , e c. I am su e I o ge someone now; I apologize and I will hank him/he
when I ealize. Many hanks also o he Neu e clinic eam o hei expe suppo wi h
language measu es in Basque and Spanish, as well as o o e ing hei clinical
pe spec i e. Special hanks o Nekane, Mai e, Amaia, and Lei e.
This sec ion I keep o colleagues ha I me along his way, and ha I am g a e ul
o call iends oday: Pie ma eo, Jo di, Gio gio, Ioanna, I ene, Chia a, Vicen e, Lau a,
Ab aham, Inés, Dani(s), Ca los, Hana, Asie , Coco. I will ne e s op eeling g a e ul o
Helena, o aking ca e o me in di icul imes. To A ai z, o le ing me b ea he and see
beyond my wo k and my ca ee , and making me unde s and ha hose a e only a (small)
pa o a li e ha can be beau i ully li ed. Many hanks o Sand a, o unde s anding
wha I needed when e en I did no know. Y a mi amilia, po acoge me con los b azos
abie os cada ez que uel o, y po hace me sen i que no ando sob e la cue da loja, y
que si ando engo una ed. All in all, hank you o keeping me close when I was
exhaus ed.
Many hanks also o all he pa icipan s ha ook pa in he s udies, and o hei
amilies. These s udies would no ha e been comple ed wi hou hem. The cu iosi y o

pa en s and child en also made me ques ion how much I knew abou wha I was
in es iga ing, so i was a uly s imula ing con ac wi h he ou side wo ld.
This hesis was suppo ed by he FPI g an BES-2016-078125 by Minis e io
Español de Economía, Indus ia y Compe i i idad MINECO and Fondo Social Eu opeo;
h ough p ojec RTI2018-096242-B-I00 (MCIU/AEI/FEDER, UE) unded by Minis e io
de Ciencia, Inno ación y Uni e sidades (MCIU), he Agencia Es a al de In es igación
(AEI) and Fondo Eu opeo de Desa ollo Regional (FEDER); by he Basque Go e nmen
h ough he BERC 2018-2021 p og am and by he Spanish S a e Resea ch Agency
h ough BCBL Se e o Ochoa excellence acc edi a ion SEV-2015-0490. The na u al
language p ocessing wo ks in S udy 1 we e suppo ed by CLARIN K-Cen e, HiTZ Cen e
- Ixa, Uni e si y o he Basque Coun y UPV/EHU.
Table o con en s
Lis o abb e ia ions ................................................................................................ ii
Resumen de la esis doc o al ................................................................................. iii
Thesis summa y ..................................................................................................... ii
Gene al in oduc ion ................................................................................................ 1
1. S udy 1. The con ibu ion o he amoun o linguis ic exposu e o bilingual
language de elopmen : Longi udinal e idence om p eschool yea s ............................. 11
1.1. In oduc ion ............................................................................................... 11
1.2. Me hod ....................................................................................................... 17
1.3. Resul s ........................................................................................................ 31
1.4. Discussion ................................................................................................. 40
2. S udy 2. The con ibu ion o ea ly language exposu e o he co ical acking
o speech: e idence om bilingual child en .................................................................... 49
2.1. In oduc ion ................................................................................................ 49
2.2. Me hod ........................................................................................................ 54
2.3. Resul s ......................................................................................................... 66
2.4. Discussion ................................................................................................... 72
3. S udy 3. Local empo al egula i ies in child-di ec ed speech ....................... 80
3.1. In oduc ion ................................................................................................. 80
3.2. Me hod ........................................................................................................ 88
3.3. Resul s ......................................................................................................... 93
3.4. Discussion ................................................................................................... 99
Gene al discussion ............................................................................................... 106
Re e ences .............................................................................................................. 117
CRediT au ho ship con ibu ion s a emen .................................................... 156
Da a a ailabili y ................................................................................................ 157
Supplemen al ma e ials ....................................................................................... 158
S udy 1 ........................................................................................................... 158
S udy 2 ........................................................................................................... 161
S udy 3 ........................................................................................................... 166
ii
Thesis summa y
The empi ical wo k ha o ms his hesis is o med by h ee in e ela ed
expe imen s (S udies 1, 2, and 3) abou he ole o linguis ic exposu e on language
de elopmen du ing childhood. Such sha ed o e a ching ques ion is add essed om
di e en empi ical pe spec i es and me hodological app oaches, and he e o e each
s udy is backed by speci ic bodies o heo e ical and empi ical esea ch. Fo his eason,
he Gene al In oduc ion o e s an o e iew o he common aspec s o he h ee s udies,
while he speci ic li e a u e ha con ex ualizes each s udy is p esen in hei espec i e
In oduc ions.
In S udy 1, we e alua ed how language de elopmen is in luenced by he amoun
o exposu e o wo languages concu en ly be ween ou and six yea s o age, by
ollowing 74 Basque-Spanish bilingual child en in h ee longi udinal s ages. Mo e
speci ically, we examined whe he di e en p opo ions o linguis ic exposu e o each
language shape language knowledge and use a phonological and non-phonological
(i.e., lexico-seman ic and syn ac ic) le els.
We obse ed a di ec impac o he amoun o linguis ic exposu e on he longi udinal
g ow h o bo h languages, especially ci cumsc ibed o lexico-seman ic and syn ac ic
abili ies. While phonological skills we e no di ec ly impac ed by exposu e, hey we e
mo e p o icien in he language wi h o e all mo e exposu e. Rega ding he de elopmen
o he spon aneous use o lexically di e se and syn ac ically ich u e ances by child en,
we obse ed ha i was sub ended by bo h he amoun o exposu e and also po en ially
by accumula ed knowledge in ela ed language domains impac ed by i (i.e., lexical and
syn ac ic knowledge). We con ex ualize ou indings in o a bilingual accoun o which

iii
language domains a e mo e pa icula ly impac ed by linguis ic inpu and how
knowledge wi hin such domains may shape he use o wo languages du ing childhood.
In S udy 2, we es ed whe he he ela i e amoun o linguis ic exposu e o each
o he languages o bilinguals (in ou case, Basque and Spanish) shapes a neu ocogni i e
abili y ele an o speech comp ehension, he co ical acking o speech. To achie e
his goal, we ollowed 35 child en om S udy 1, namely hose who had he bigges
exposu e o Basque (L1) and smalles o Spanish (L2), in o de o es whe he such
unbalanced accumula ed inpu in each language played a ole in he ma u a ion o he
co ical acking o phonological and non-phonological dis ibu ional p ope ies o
speech, o each language sepa a ely. We did no ind e idence o a di ec in luence o
he amoun o exposu e o he ma u a ion o co ical acking o speech. Howe e , we
obse ed speci ic ela ionships be ween he co ical acking o speech a phonological
and non-phonological le els and beha io al pe o mance in he espec i e domains,
al hough his inding was ci cumsc ibed o he dominan language o child en (L1,
Basque). We expec ou indings o in o m u u e de elopmen al esea ch abou he
ac o s in luencing he de elopmen o he co ical acking o speech, as well as how
his neu al mechanism con ibu es o speech abili ies measu ed in child en.
In S udy 3, we shi ed he ocus o ou in es iga ion om quan i a i e (amoun
o exposu e) o quali a i e aspec s o he linguis ic inpu ha child en a e exposed o.
We es ed 18 women, na i e speake s o Spanish, while spon aneously add essing hei
child en (child-di ec ed speech), o some adul s (adul -di ec ed speech). We used a
spec o- empo al model and es ima ed he syllable a e, he p osodic p ominence, and
he egula i y wi h which syllables a e s essed in child-di ec ed speech in compa ison
ix
o adul -di ec ed speech. We obse ed ha child-di ec ed speech was cha ac e ized by
a slowe syllable a e, mo e p osodic salience, as well as mo e egula alignmen
be ween p osodic and syllabic ampli ude modula ions han adul -di ec ed speech. Ou
indings a e in line wi h cu en heo ies abou he eme gence o phonological abili ies,
by highligh ing empo al egula i ies in child-di ec ed speech ha could be exploi ed
by a de eloping phonological sys em. These obse a ions a e also ela ed o ou indings
in S udy 2 abou he co ical acking o speech, and can se e a mul ile el pe spec i e
o phonological de elopmen .
1
Gene al in oduc ion
The ole o o al linguis ic exposu e on language de elopmen du ing
childhood
In e e yday li e, we encoun e speech in di e en o ms and in a as numbe o
con ex s, wha makes indi iduals become p o icien in hei languages wi h an appa en
ease h oughou de elopmen . None heless, a g ea amoun o in e nal and
en i onmen al ac o s will modula e he a e and ease wi h which indi iduals achie e
such p o iciency. Among indi idual-le el de e minan s o language acquisi ion (also
ele an o b oade cogni i e de elopmen ), ch onological age and “sensi i e pe iods”
(We ke & Hensch, 2015) a e c ucial o a aining language miles ones, such as he
consolida ion o language-speci ic phonemic ca ego ies (e.g., Bu ns e al., 2007) and he
p oduc ion o i s wo ds (see Kuhl, 2004). A delay in eaching such miles ones migh
hampe he de elopmen o a ious linguis ic p ocessing skills (i.e., phonological,
lexical, syn ac ic; e.g., Banai & Ahissa , 2018; Bishop & Snowling, 2004; Ni oue &
Bu on, 2005), esul ing in pe o mance lags wi h espec o ypically de eloping pee s,
such as wha is seen in neu ode elopmen al language diso de s (e.g., de elopmen al
dyslexia, de elopmen al language diso de ).
Pa allel o indi idual-le el aspec s, con ex ual (o en i onmen al) ac o s will
ei he posi i ely o nega i ely in luence language de elopmen depending on he
ichness o he inpu s o which young lea ne s a e exposed. The main ques ion o he
p esen hesis is speci ically aimed a explo ing he ole o he ichness o o al linguis ic
inpu on language de elopmen du ing childhood, h ough he s udy o he quan i y
and he quali y o exposu e o a language.
2
A c ucial sou ce o in e -indi idual a ia ions modula ing language de elopmen
ajec o ies is he amoun o exposu e o a language, e lec ing quan i a i e aspec s o
he speech inpu (e.g., Pea son e al., 1997; Tho da do i , 2011). Bilingualism (as well as
mul ilingualism) o e s he oppo uni y o explo e he ole o he amoun o exposu e
on language acquisi ion h ough he measu emen o he ela i e inpu in each
language. S udies 1 and 2 will ocus on es ima ing he di ec in luence o he amoun o
linguis ic exposu e o languages on he de elopmen o knowledge o hese languages
in bilingual child en, a bo h he beha io al and neu ocogni i e le els.
In addi ion o he amoun o exposu e o a language, he quali y o he linguis ic
inpu impac s language de elopmen o a conside able ex en . Fo example he
socioeconomic s a us o he indi idual and hei household is a sizable and pe asi e
con ex ual ac o which has a s ong impac on language acquisi ion (Dailey &
Be gelson, 2022; Ha & Risley, 1995; Ho , 2003a; Me z e al., 2020; Pace e al., 2017).
O pa icula ele ance o his hesis, one o he mos widely s udied media o s be ween
socioeconomic s a us and language de elopmen is he speech egis e used by adul s
o communica e wi h young child en, known as child-di ec ed speech (C is ia, 2022;
Dailey & Be gelson, 2022; Ma chman e al., 2017). Child-di ec ed speech appea s o
boos language lea ning h ough he adap a ion o spec al (e.g., pi ch) and empo al
ea u es (e.g., speech a e), which engage and ease i s p ocessing (Fe nald, 2000; Leong
& Goswami, 2015). The e o e, he in es iga ion o child-di ec ed speech as a window
in o language de elopmen o e s ways o de e mine he speci ic cha ac e is ics o he
speech signal ha may p o ide lea ning- acili a o y e ec s. Fo his eason, he s udy o
child-di ec ed speech will be he speci ic ocus o S udy 3.

3
Language knowledge ac oss undamen al domains
Wi h espec o he de elopmen o language abili ies, se e al concep s will be
ackled by he h ee s udies o he hesis, and he e o e we a e de ining hem below.
Language knowledge (also e e ed o in some occasions as p o iciency) is he
gene al concep wi h which we app oxima e language de elopmen , and can be hough
o as he capaci y o comp ehend and p oduce a gi en language. We will di e en ia e
wo b oad ca ego ies o language knowledge: phonological and non-phonological
abili ies. Phonological abili ies suppo he sensi i i y, ca ego iza ion, and
manipula ion o he sounds o a language, as well as he ules o dis ibu ional
p ope ies ha go e n hose sounds. Consequen ly, non-phonological abili ies a e hose
language domains ha do no di ec ly ela e o he ca ego iza ion o speech sounds, bu
a he o he knowledge highe o highe -le el linguis ic uni s (i.e., wo ds and
sen ences). In his hesis, we ocus on wo non-phonological domains: lexico-seman ic
abili ies, ha allow o he comp ehension and use o ocabula y; and syn ac ic skills,
which enable lea ne s o ex ac he speci ic meaning o a linguis ic s uc u e (i.e., a
sen ence o a g oup o sen ences), and h ough which we app oxima e b oade
g amma ical knowledge.
The e o e, language knowledge wi hin he men ioned domains will ac as a
dependen a iable o he (quan i y and quali y o ) linguis ic exposu e.
Quan i y and quali y o linguis ic exposu e on language de elopmen
Quan i a i e exposu e o a language could be de ined as he passi e con ac wi h
linguis ic inpu s by he me e p esence o he indi idual in he con ex in which i is
4
used. He e, we concep ualize he amoun o exposu e o a language as a p ocess h ough
which he child is no only a passi e lis ene , bu also ac i ely uses and p oduces his
language. Such dis inc ion is ele an because depending on whe he he linguis ic
inpu , o he inpu and i s use (as in ou case), a e assessed, di e en language ou comes
can be obse ed (Bedo e e al., 2012). Thus, he de ini ion o amoun o exposu e
(AoE) in he p esen hesis e e s o he ime spen wi hin a con ex in which a gi en
language is spoken. Such de ini ion allows us o con ex ualize ou in es iga ion among
esea ch ha used a simila e m o explo e he ole o linguis ic inpu in language
de elopmen , including s udies in bilingual and monolingual child en, child en wi h
hea ing impai men s, and child en g owing in di e se socioeconomic and cul u al
con ex s (e.g., C is ia e al., 2019; Ni oue & Bu on, 2005; Pea son e al., 1997). He e,
we s udy bilingualism as a window in o he impac o he amoun o linguis ic exposu e
on language de elopmen . Indeed, he quan i ica ion o AoE in each o he languages o
bilinguals o e s he oppo uni y o di ec ly assessing he e ec o di e en p opo ions
o exposu e o wo languages wi hin he same indi iduals.
In addi ion o AoE, se e al ac o s di ec ly linked o bilingualism ha e been
shown o play a ole in language de elopmen such as he age o acquisi ion, con ex s
o language use (dual e sus single-language con ex s), socio-cul u al ac o s o use
(e.g., he i age languages, majo i y and mino i y languages), e c. Toge he , hese
di e en ac o s o e a ai h ul pic u e o he con ex and his o y o language lea ning
ha can explain he le el o p o iciency in di e en linguis ic domains. Impo an ly, a
ecen Delphi consensus o esea che s, eache s, and speech and language he apis s
ag eed ha a comp ehensi e s udy o bilingualism should include measu es o
“language exposu e and use as well as he assessmen o language impai men s,
5
p o iciency le els, educa ion and li e acy his o y, he quali y o he inpu , language
mixing p ac ices, and a i udes owa ds languages and language mixing” (Ca e al.,
2022). Such sub ac o s o bilingualism o e a comp ehensi e accoun o he con ex in
which bilingual de elopmen akes place. Among hose ac o s, we ocus on how AoE
o wo languages shapes hei de elopmen .
Mo e speci ically, we quan i y indi idual AoE o he wo languages o bilingual
child en as a con inuous a iable, o explo e i s di ec impac on he de elopmen o
language knowledge. Such con inuous AoE index allows us o o e come some o he
disad an ages linked o he s udy o exposu e h ough less di ec measu es based on
be ween-g oups compa isons (bilinguals e sus monolinguals) o o he bilingual-
speci ic ea u es (e.g., L1 s. L2). Impo an ly, ou exposu e measu e allows us o explo e
he e ec o e y limi ed amoun s o exposu e (e.g., less han 30% o he ime in he
case o L2) on language de elopmen in child en “becoming bilinguals”. In addi ion, i
has been ecen ly alida ed by me hodological in es iga ions ha show ha a
mul i ac o ial and con inuous cha ac e iza ions o bilingual exposu e accu a ely
p edic s language pe o mance (Gulli e e al., 2020; Gulli e & Ti one, 2019).
In he i s wo s udies, we capi alized on such con inuous measu e o explo e
he ole o quan i y o exposu e on bilingual language de elopmen . In he hi d s udy,
we explo ed linguis ic exposu e om a complemen a y pe spec i e, namely looking a
he quali y o he inpu o which he child en a e exposed o.
He e we de ine quali y o exposu e as he a ailabili y o enhancemen o a se
o speech ea u es ha acili a es i s e icien analysis by a language-lea ning indi idual.
One example o he ele ance o he quali y o exposu e o language acquisi ion and
6
de elopmen is po ayed by s udies in child en wi h hea ing impai men s, whose
impo e ished pe cep ion o acous ic s imuli esul s in delays ac oss di e en linguis ic
domains (Ni oue , 1996; Ni oue & Bu on, 2005). In addi ion, i has been shown
ha he quali y o he speech used by pa en s o communica e o hei child en, i.e.,
adap ed, di e se, and inc easingly complex in e ms lexical and syn ac ic p oduc ions,
posi i ely in luences language de elopmen (Fu ow e al., 1979; Ho -Ginsbe g, 1986;
Hu enloche e al., 2002, 2010). And he las example, o pa icula ele ance o his
hesis, is he adap a ion o spec al (i.e., pi ch, one) and empo al (e.g., speech a e)
ea u es o speech ha adul s p oduce when speaking o in an s and child en. In an -
di ec ed speech egis e s ha e dis inc i e acous ic cha ac e is ics om adul -di ec ed
speech, which could make hem especially sui able o lea ne s o whom language
p ocessing is s ill a complex and demanding ask (Fe nald, 2000; Leong e al., 2017).
Thus, he pu pose o S udy 3 is o assess whe he , beyond in an -di ec ed speech, child-
di ec ed speech also p o ides child en wi h enhanced empo al s a is ics ha migh
suppo phonological de elopmen .
In summa y, we ope a ionalize quan i y (p opo ional amoun o bilingual
exposu e, o AoE) and quali y ( empo al cha ac e is ics o child-di ec ed speech) o o al
linguis ic inpu in o de o es ima e hei impac on he language de elopmen o
child en ha , while ha ing al eady acqui ed a conside able amoun o knowledge, a e
no ye p o icien language use s.
O al linguis ic exposu e in child en’s language de elopmen
Ou a iables o in e es o s udy he ole o quan i a i e and quali a i e ichness
o he speech inpu in language de elopmen a e de ined based on he o al na u e o he
13
pe sis en when he inpu is limi ed, as in bilingual de elopmen (Gibson e al., 2012;
Olle e al., 2007; Windso & Kohne , 2004). Such ecep i e-exp essi e di e ence
poin s a a po en ially s onge ela ionship be ween AoE and ecep i e han p oduc i e
ocabula y. Cohe en wi h he “ ecep i e-exp essi e gap” hypo hesis, Gigue e and Ho
(2022) ound ha ini ial di e ences in ecep i e ocabula y be ween L1 and L2
diminished be ween 4.5 and 10 yea s o age, while he L1-L2 gap was pe sis en in
exp essi e ocabula y. Toge he , he o me s udies show ha child en e en could s ill
bene i om scan AoE o de elop hei ocabula y comp ehension abili ies in hei
non-dominan language. In summa y, ocabula y lea ning could be hough o as a
“da a-hung y” p ocess (C is ia, 2020) which eeds om all he —e en limi ed— inpu
a ailable, bu leading o bigge gains in he ecep i e han he p oduc i e subdomain.
While he ela ionship be ween AoE and lexical de elopmen is s aigh o wa d
as i akes place on an i em-based ashion, he in luence o AoE on syn ac ic
de elopmen is no as di ec ly obse able. The ac ha syn ac ic s uc u es a e
combina o ial makes hem less eadily epea ed in he linguis ic inpu han wo ds,
which appea s o limi he di ec ole o AoE on hei de elopmen when compa ed o
lexical abili ies (Olle e al., 2007; Pa adis & Genesee, 1996). S ill, i has also been
obse ed ha bilingual child en can show educed le els o bo h lexical and syn ac ic
knowledge compa ed o monolinguals (Chond ogianni & Ma inis, 2011; Tho da do i
e al., 2006). The e o e, i is s ill needed o de e mine whe he and o wha ex en
language-speci ic AoE cons ains bilingual syn ac ic knowledge de elopmen . In a o
o he ele ance o AoE o syn ac ic knowledge de elopmen , De Houwe (2005)
showed ha such de elopmen in bilinguals migh occu independen ly o each
language —hence depending on language-speci ic ac o s such as AoE— and ollow a

14
ajec o y simila o wha is obse ed in monolinguals. In monolinguals, i has been
consis en ly shown ha ca egi e s’ use o inc easingly complex syn ac ic s uc u es
p edic s hei child en’s syn ac ic de elopmen (Fu ow e al., 1979; Ho -Ginsbe g,
1986; Hu enloche e al., 2002, 2010). In bilinguals, Yip and Ma hews (2006) ound
ha he g ow h o he mean leng h o u e ance (MLU), a classical p oxy o syn ac ic
de elopmen , s ongly depends on language dominance: while L1 and L2 MLU g ew
s eadily be ween 2 and 3 yea s o age in hei s udy, L1 MLU was consis en ly bigge han
in L2. Howe e , o he bes o ou knowledge, he ole o AoE on language-speci ic
bilingual syn ac ic de elopmen is s ill an open ques ion ha he p esen s udy will
di ec ly add ess.
Unlike in lexical and syn ac ic de elopmen , AoE could play a mino ole in
phonological de elopmen (see C is ia, 2020 o an o e iew). This seems due o he
ac ha he acquisi ion o phonological abili ies elies mo e on he i able ac o s han
he o me domains (Bishop, 2002; Ko as e al., 2005), which could make i s ypical
de elopmen de ached om exposu e o a conside able ex en . Indeed, because lexico-
seman ics, syn ax and phonology migh play di e en oles du ing language acquisi ion,
some au ho s ha e p oposed o di ide he ela ed skills in “phonological” and “non-
phonological” (i.e., lexico-seman ic and syn ac ic skills) abili ies (Bishop & Snowling,
2004). This di ision has p o en use ul o di e en ia e he po en ial causes o language-
ela ed de elopmen al diso de s such as de elopmen al dyslexia ( ha migh occu
because o phonological de ici s) and de elopmen al language diso de ( ha migh
esul om bo h phonological and non-phonological de ici s). None heless, linguis ic
inpu cha ac e is ics should in luence a leas o some ex en , he de elopmen o
language-speci ic phonological abili ies. Such in luence is illus a ed by he b oadly
15
s udied impac o bilingualism (as opposed o monolingualism) on he de elopmen o
phonemic ca ego ies in in ancy and ea ly childhood, showing a simila (al hough
sligh ly delayed in some cases) de elopmen o phonemic bounda ies in each language
(Bosch & Sebas ián-Gallés, 2003; Bu ns e al., 2007; Kuhl e al., 1997; Ruiz-Fel e e al.,
2016). Howe e , ewe s udies hin a he po en ial ole o AoE on he de elopmen o
phonological capaci ies du ing childhood (e.g., Gu ié ez-Clellen & Simon-Ce eijido,
2010; Messe e al., 2010; Summe s e al., 2010). In e es ingly, Pa a, Ho , and Co e
(2011) ound ha , in 2- o 3-yea -old child en, AoE o L1 and L2 p edic ed he epe i ion
o nonwo ds wi h phonemic and syllabic phono ac ic p ope ies o each language
espec i ely. In he same s udy, i was ound ha nonwo d epe i ion pe o mance was
posi i ely co ela ed be ween bo h languages, which poin s a bo h inpu -dependen
and inpu -independen ac o s in phonological de elopmen . In addi ion, Pa a e al.
(2011) epo ed ha AoE p edic ed ocabula y mo e han phonology, sugges ing a
po en ial g ea e ole o AoE o he de elopmen o non-phonological compa ed o
phonological abili ies. Howe e , his hypo hesis mus also accoun o indings showing
ha phonological abili ies a e ele an p edic o s o non-phonological language abili ies
such as ocabula y knowledge (An hony e al., 2003; Ca oll e al., 2003; Kehoe e al.,
2020; Vaah o an a e al., 2020). In ac , i is easonable o assume ha mo e e icien
and p ecise coding, s o age, and e ie al p ocesses o speech sounds (i.e., phonological
p ocessing) should con ibu e o build up s able lexical ep esen a ions.
O e all, p e ious s udies ha e o e ed mul iple snapsho s o he in luence o
language use and AoE on he de elopmen al ajec o y o di e en language domains
du ing ea ly childhood. They sugges ha AoE migh play a s onge ole in he
de elopmen o non-phonological (lexico-seman ics and syn ax) compa ed o
16
phonological abili ies. In he p esen s udy, we aim o assess he ole o AoE on bo h
phonological (phonological awa eness and phonological sho - e m memo y) and non-
phonological (lexico-seman ic and syn ac ic) de elopmen in child en a h ee s ages
spanning o e p eschool yea s and he i s yea o p ima y school (4 o 6 yea s o age).
Impo an ly, we capi alize on AoE a ia ions be ween he wo languages o ea ly
Basque-Spanish bilingual child en o explo e hese e ec s in a no el way. By es ing AoE
and language knowledge and use longi udinally o each language h ough a wi hin-
subjec design, we plan o d aw a comp ehensi e pic u e o he ole o AoE in (bilingual)
language de elopmen . The mo e speci ic aim o his s udy was wo old.
Fi s , we wan ed o de e mine whe he AoE has an impac on he de elopmen al
ajec o y o he classical gap obse ed be ween pe o mance in he i s and he second
language o bilinguals (e.g., Gigue e & Ho , 2022; Haman e al., 2017; Pa adis e al.,
2016). As AoE was dominan in Basque o e Spanish in ou g oup o pa icipan s
1
, we
expec ed an ini ially be e g oup-le el pe o mance in Basque han Spanish ac oss
language domains. We expec ed di e en de elopmen al ajec o ies be ween bo h
languages, dependen also on he linguis ic domain s udied (i.e., lexico-seman ics,
syn ax, and phonology). In pa icula , we expec ed non-phonological abili ies o show
a signi ican gap be ween Basque and Spanish, especially in he case o lexical abili ies
as hey ha e been shown o be mo e in luenced by inpu han syn ac ic abili ies (Olle
e al., 2007; Pa adis & Genesee, 1996). On he o he hand, we did no expec signi ican
1
Since pa icipan s wi h di e en p opo ions o exposu e o Basque and Spanish we e included,
we canno e m hese languages as L1 and L2 espec i ely because bo h languages we e lea ned
simul aneously since ea ly in li e and a small g oup o pa icipan s had mo e ela i e exposu e o Spanish
han Basque.
17
di e ences in phonological abili ies be ween Basque and Spanish a he s a ing poin
o he s udy no in hei de elopmen al ajec o ies.
Second, we aimed o shed ligh on o whe he he in luence o AoE h oughou
he de elopmen al ime span o he s udy (i.e., aking in o accoun he h ee
longi udinal ime poin s al oge he ) is speci ic o non-phonological skills, i.e., lexico-
seman ics and syn ax, o whe he i is s able ac oss language domains including
phonological skills, oo. We p edic ed ha he in luence o AoE on language
pe o mance would be posi i e o lexical and syn ac ic abili ies, bu no (o less so) o
phonological skills. Howe e , we did no ha e a clea p edic ion abou changes a ec ing
he ela ionship be ween AoE and language pe o mance wi hin each de elopmen al
s age (i.e., each longi udinal es ing poin ).
1.2. Me hod
1.2.1. Pa icipan s
Se en y- ou child en (36 emales) ha g ew up in a Basque-Spanish bilingual
con ex we e ec ui ed o ake pa in a longi udinal assessmen . They we e es ed h ee
imes (T1, T2 and T3) spanning o e he las yea o p eschool and he i s yea o
p ima y school: T1 – 4.01 yea s old (SD = .07), T2 – 4.71 y.o. (SD = .11), and T3 – 6.4 y.o.
(SD = .1). Mo e de ailed in o ma ion abou how pa icipan ’s amoun o exposu e was
assessed and abou linguis ic exposu e pa e ns is desc ibed below (subsec ions Amoun
o exposu e and G oup cha ac e is ics o bilingual exposu e). Child en’s pa icipa ion
was ewa ded wi h an educa ional gi , and hei pa en s we e in o med abou he
esea ch aims and ou comes once a he end o he s udy. Pa icipan ec ui men
18
p ocedu e and da a collec ion pa adigms we e app o ed by he BCBL E hics Commi ee
and complied wi h he Decla a ion o Helsinki. Th ee child en we e excluded om he
longi udinal es ing as hei exposu e o Basque was minimum (i.e., i ually none) and
hus we e no able o comple e he asks wi h he minimum pe o mance equi ed in
each language o ou goals.
1.2.2. Longi udinal assessmen
A each s age, he assessmen included (i) he adminis a ion o a ques ionnai e
o he pa en s/legal u o s o pa icipan s ha was aimed a cha ac e izing he linguis ic
backg ound o hei child, and (ii) an in-lab e alua ion. Wi hin his in-lab e alua ion,
we assessed child en’s pe o mance in lexico-seman ics, syn ac ic and phonological
abili ies in Basque and Spanish. I was composed o a se ies o compu e ized
“con olled” asks ( o which he s imuli selec ion and p esen a ion we e con olled and
balanced be ween he Basque and Spanish e sions), and o a ask c ea ed o elici a
spon aneous con e sa ion be ween he child and one o hei pa en s. The objec i e o
his las ask was o assess child en’s language p oduc ion skills in each language in an
ecologically alid con ex simila o hei e e yday li e. In addi ion o he language
measu es, we assessed child en’s non-linguis ic cogni i e abili ies wi h he ma ices
sub es o Kau man B ie In elligence Tes (KBIT-2, Kau man & Kau man, 2014).
Howe e , he as majo i y o pa icipan s pe o med a ceiling le el ela i e o hei age
s anda ds and hus KBIT-2 measu es did no o e enough a iabili y o be included in
he s a is ical models.

19
Amoun o linguis ic exposu e - AoE
We es ima ed he amoun o exposu e (AoE) o each language h ough a
mul i ac o ial index p e iously alida ed by o he s udies o bilingual language
de elopmen (Gulli e e al., 2020; Tho da do i , 2011; Tho da do i e al., 2006). The
pa en s o he child en pa icipan s p o ided in o ma ion on hei child en’s language
exposu e and use since bi h h ough an ex ensi e online ques ionnai e (Amoun o
exposu e ques ionnai e, a ailable a he Open Science F amewo k eposi o y o he
p ojec ). We used his in o ma ion o build composi e indexes o AoE o bo h Basque
and Spanish. The composi e AoE indexes comp ised he age o acquisi ion o each
language, he pe cen age o waking hou s ha a child had been exposed o each
language be ween bi h and each longi udinal s age, as well as a de ailed cu en AoE
o hei languages a school, home, and leisu e con ex s. An AoE index o 100 % in a
language would indica e exposu e o a unique language since bi h, while 0 % would
indica e no con ac wi h ha language wha soe e . The speci ic ope a ionaliza ion o
he AoE indexes can be ound in he Supplemen a y Ma e ials (Supplemen al Fo mulas
1.1, 1.2, and 1.3). Ou AoE me ic can be concep ualized as a measu e ha pu s oge he
he “day-in- he-li e” in e iew (e.g., Res epo, 1998) and he cumula i e li e his o y o
language exposu e since bi h (e.g., Tho da do i e al., 2006), which ha e p o en
e ec i e me hods o cap u ing ea ly bilingual expe ience and a e ela ed o language
knowledge ac oss he domains o in e es (phonology, lexico-seman ics, and syn ax) o
he p esen s udy (e.g., Gámez e al., 2019; Tho da do i , 2011; Tho da do i e al.,
2006). Mo eo e , all he obse a ions aken in o conside a ion o build ou AoE sco es
comply wi h he ecen ecommenda ions o he Delphi consensus o esea che s and
he apis s on p o ocols o cha ac e ize child en’s bilingual exposu e (Ca e al., 2022).
20
G oup cha ac e is ics o bilingual exposu e
Mos pa icipan s (and he e o e he g oup as a whole) we e simul aneous
Basque-Spanish bilinguals, as hey had s a ed acqui ing bo h languages a oughly he
same ime, as well as Basque-dominan , gi en ha hey we e signi ican ly mo e exposed
o Basque han Spanish (see Age o acquisi ion in Table 1.1., and % o exposu e in he las
panel o Figu e 1.1). In he Basque coun y, such exposu e p o ile is ypical as he
sociocul u al and educa ional policies p omo e he ea ly acquisi ion o Basque which is
a mino i y language now in ac i e eco e y (see Zalbide & Cenoz, 2008). In ou g oup,
Spanish ( he majo i y language in he Basque Coun y) was ypically acqui ed as a
second language. Such di e ences in exposu e be ween Basque and Spanish, in addi ion
o a con inuous ange o AoE o each language espec i ely, we e key o add essing ou
esea ch ques ions abou he ole ha AoE, and speci ic sus ained AoE di e ences, play
on (bilingual) language de elopmen ac oss undamen al language domains.
S age
Sample
Age
% AoE
Basque
% AoE
Spanish
AoA
Basque
AoA
Spanish
1
71 (35 em.)
4.01 (.07)
66.81 (22.69)
25.17 (22.11)
0.22 (.67)
0.74 (1.11)
2
65 (30 em.)
4.71 (.11)
71.53 (20.06)
23.92 (22.37)
3
63 (30 em.)
6.4 (.1)
71.08 (21.77)
26.14 (22.36)
Table 1.1. Gene al in o ma ion abou sample size and a e age age (in yea s), pe cen age o
exposu e (% AoE) and age o acquisi ion (AoA, in yea s) in each language. S anda d de ia ion
o each measu e is ep esen ed be ween pa en heses.
Lexico-seman ics
21
Lexico-seman ics was assessed by means o h ee measu es: a p oduc i e and a
ecep i e ocabula y measu e ia, espec i ely, a compu e ized pic u e-naming and
a wo d-comp ehension ask in each language, as well as a lexical di e si y index
ex ac ed om he spon aneous speech p oduc ions o child en when alking wi h hei
pa en s.
P oduc i e and he ecep i e ocabula y asks
Fo bo h asks, he numbe and di icul y o i ems inc eased along he h ee
longi udinal s ages. Thus, pic u e-naming ask consis ed o 23 i ems in he 1s s age o
he s udy, and o 45 i ems in he 2nd and 3 d s ages; in wo d-comp ehension ask,
pa icipan s we e p esen ed wi h 19 i ems in he 1s s age, and 25 in he 2nd and 3 d s ages.
Rega ding he di icul y o he i ems, we selec ed a ge wo ds wi h dec easing lexical
Zip equency (a p oxy o wo d di icul y, see an Heu en e al., 2014) along he s ages
o he s udy o p oduc i e (Basque: s age 1 mean Zip equency = 3.90 (SD = 0.62), s age
2 = 3.65 (SD = 0.59), s age 3 = 3.62 (SD = 0.54); Spanish: s age 1 = 4.25 (SD = 0.47), s age
2 = 4.04 (SD = 0.44), s age 3 = 3.94 (SD = 0.57)) and ecep i e ocabula y (Basque: s age
1 mean Zip equency = 3.64 (SD = 0.71), s age 2 = 3.58 (SD = 0.49), s age 3 = 2.97 (SD =
.57); Spanish: s age 1 = 4.17 (SD = .65), s age 2 = 3.94 (SD = .37), s age 3 = 3.32 (SD = .33))
espec i ely. In he pic u e-naming ask, pa icipan s we e equi ed o name he
pic u es ha appea ed on sc een in a andomized o de , wi hou any ime cons ain .
In wo d-comp ehension ask, pa icipan s we e p esen ed wi h an audi o y wo d o e
speake s, and p esen ed wi h ou pic u es on sc een a he same ime. They we e asked
o poin a he pic u e co esponding o he wo d ha hey hea d. The h ee o he
pic u es co esponded o dis ac o s ha p esen ed ei he phonological, isual, o
22
seman ic simila i ies wi h he a ge wo d. The esea che coded each ial as ei he
co ec o inco ec upon esponse. The o de o p esen a ion o he wo languages in
bo h o he ocabula y asks was coun e balanced ac oss pa icipan s. In o de o
ha monize pa icipan s’ pe o mance in ocabula y asks ac oss languages and s ages,
we weigh ed pe o mance in he di e en i ems (1, co ec ; 0, inco ec ) by he in e se
o hei Zip lexical equency (ex ac ed om EHME da abase o Basque wo ds, Acha
e al., 2014; and om EsPal da abase in he case o Spanish, Duchon e al., 2013).
The e o e, mo e equen (and easie ) wo ds had a smalle weigh on pa icipan s’
o e all pe o mance han less equen ones ha we e less likely o be known by
child en (see Supplemen al Fo mula 1.4). Thus, ou measu e o pe o mance in pic u e-
naming and wo d-comp ehension asks was Zip - equency-weigh ed accu acy.
Lexical di e si y
The hi d o ou lexico-seman ic measu es, lexical di e si y, was ex ac ed om
a con e sa ional co pus o spon aneous speech p oduc ions, ha also se ed o
es ima ing na u alis ic indexes o syn ac ic abili ies (namely, clausal densi y and mean
leng h o u e ance, desc ibed below wi hin he syn ac ic abili ies subsec ion). The
se ing o ha ask was aimed o be simila o he home en i onmen , so ha he child
and hei pa en could in e ac eely, like hey would do i hey we e playing wi h
pic u e books, oys, o games a home. We in o med pa en s o he pu pose o he ask,
and p o ided hem wi h he ollowing simple ins uc ions: “please, in e ac wi h you
child as you would do a home, while allowing he /him o alk as much and as na u ally
as possible in hei p e e ed language(s).” P e ious s udies epo ed ha 7–10 minu e-
long eco dings a e su icien o ob ain eliable es ima es o language p o iciency in
29
1.2.3. Da a analysis
P io o modeling he in luence o AoE on he longi udinal ajec o y o bilingual
language ou comes, we de ec ed and emo ed ou lie s based on he in e qua ile ange
(IQR) c i e ion. This way, we emo ed om each o he a iables o in e es da apoin s
ha we e o e 1.5 imes he in e qua ile ange abo e he 75 h pe cen ile o unde 1.5
imes he in e qua ile ange below he 25 h pe cen ile.
The main goal o ou longi udinal s udy was o es ima e bilingual language
de elopmen ac oss he h ee language domains as a unc ion o AoE and ime (i.e.,
longi udinal s age). In pa icula wo ques ions we e in es iga ed ela ed o: (i) he
di e ences in longi udinal g ow h ajec o ies o unbalanced bilinguals be ween hei
wo languages, aken as a “g oup-le el” p oxy o AoE; (ii) he longi udinal in luence o
AoE on language pe o mance, h ough he s udy o child-speci ic con inuous indexes
o AoE. To add ess each ques ion comp ehensi ely, wo se s o analyses we e pe o med
(see below). Conduc ing hese wo analyses on he same da a was o simpli y he
in e p e a ion o ou esul s, by no ha ing mul iple in e ac ions be ween ca ego ical
(i.e., language) and con inuous ac o s (i.e., AoE).
In o de cap u e he mul ile el s uc u e o ou da a, consis ing o epea ed
wi hin-pa icipan obse a ions ac oss di e en domains, languages, and longi udinal
s ages, we used pa icipan s as andom in e cep s h oughou all LME models, which
allowed us o accoun o baseline indi idual di e ences (i.e., one in e cep pe
pa icipan and model) in he di e en languages and language domains.
We used lme o mula om lme4 package (Ba es e al., 2015) in R ( e sion 4.2.1,
R Co e Team, 2022) o i he di e en LMEs. In o de o es o omnibus main e ec s

30
and in e ac ions o he p edic o s, we used he ano a unc ion o base R; and be ween-
language and be ween-longi udinal s age di e ences in he each model we e e alua ed
wi h di lsmeans om he lme Tes package (Kuzne so a e al., 2017).
1.2.3.1. E ec o language (Basque s. Spanish) as a g oup-le el p oxy o AoE
Since he g oup-le el AoE was signi ican ly di e en be ween Basque and
Spanish (i.e., Basque-dominan ), we ope a ionalized he ac o “language” as a p oxy o
AoE (Basque AoE > Spanish AoE). I is ele an o no e ha language is only an
app oxima ion o language dominance a g oup le el and, as such, is in luenced by o he
ac o s in addi ion o AoE (such as age o acquisi ion). To assess he e ec o language
on longi udinal g ow h a each de elopmen al s age, we i ed one linea mixed e ec
(LME) model o each collec ed measu e o language pe o mance as ou come a iable,
wi h s age (1, 2, 3) and language (Basque, Spanish), as well as hei in e ac ion, as ixed
e ec s.
1.2.3.2. Di ec e ec o child-speci ic AoE indexes on each domain
Al hough es ing he e ec o language on he longi udinal ajec o y o he
di e en measu es is use ul o in es iga e language dominance-dependen di e ences
in he de elopmen al cou se o a language measu e ( he a o emen ioned i s se o
analyses), assessing how indexes di ec ly measu ing AoE in luence pe o mance is key
o unde s and he di ec impac o AoE on he de elopmen o each linguis ic domain,
in each language. Fo his second se o analyses, we buil wo addi ional subse s o LME
models wi h he same s uc u es o each language pe o mance measu e. The objec i e
o he i s subse o models was o assess he o e all e ec o AoE on language
31
de elopmen and whe he he e we e be ween-languages di e ences in he in luence o
AoE on language abili ies independen ly o de elopmen al s age. To achie e his, o each
measu e as he ou come a iable, we i ed an LME model wi h he con inuous child-
speci ic AoE (in %), and he in e ac ion be ween hese indi idual AoE indexes and
language (Basque, Spanish) as ixed e ec s. The aim o he second subse o models was
o del e deepe in o he speci ic ela ionship be ween indi idual AoE indexes and
language pe o mance as a unc ion o de elopmen al s age, wi hin Basque and Spanish
sepa a ely ( he e ec o language being assessed in he i s se o models, i.e., 1.2.3.1).
Thus, we i ed wo LME models, one in Basque and one in Spanish, wi h he in e ac ion
be ween indi idual AoE indexes and s age as ixed e ec s.
1.3. Resul s
1.3.1. Longi udinal g ow h ac oss domains in each language
Lexico-seman ic abili ies
Fi s , we es ed he longi udinal ajec o y o lexico-seman ic abili ies. LME
yielded a signi ican e ec o language, F(1, 327.39) = 60.65, p < .001, and s age, F(2,
334.37) = 83.77, p < .001, on p oduc i e ocabula y, and no signi ican in e ac ion
be ween bo h ac o s, F(2, 326.62) = 1.52, p > .05. This is isible in Figu e 1.1, which
shows o e all highe p oduc i e ocabula y in Basque han in Spanish ac oss s ages
(s age 1: (326.8) = 6.06, p < .001 (β = .055, SE = .009, CI [.037 .072]; s age 2: (326.5) =
3.47, p < .001 (β = 0.033, SE = .009, CI [.014 0.051]); s age 3: (327.3) = 4.04, p < .001 (β
= 0.039, SE = .010, CI [.02 0.058])). The e was also a pa allel longi udinal g ow h o
32
bo h languages wi h no signi ican inc ease be ween he 1s and 2nd s ages in Basque,
(330.3) = -0.28, p > .05 (β = -0.003, SE = .009, CI [-0.021 0.015]), and only a small
signi ican inc ease in Spanish, (332) = 2.1, p = .037 (β = .02, SE = .009, CI [0.012
0.038]); while be ween he 2nd and 3 d s ages he e was a la ge signi ican inc ease (o
simila magni ude) in bo h languages (Basque: (328) = 7.75, p < .001 (β = .073, SE =
.009, CI [0.055 0.092]); Spanish: (327.3) = 7.07, p < .001 (β = .067, SE = .01, CI [0.049
0.086])).
Wi h espec o ecep i e ocabula y, he e we e signi ican main e ec s o
language, F(1, 327.3) = 77.23, p < .001, s age, F(2, 334.85) = 158.91, p < .001, on wo d
comp ehension pe o mance, as well as an in e ac ion be ween language and s age, F(2,
325.97) = 60.65, p < .001. Such in e ac ion was he p oduc o Spanish ecep i e
ocabula y g owing a a s eadie and as e a e han Basque, as isible in Figu e 1.1.
While he di e ence be ween he 1s and 2nd s ages in Basque ecep i e ocabula y was
no signi ican , (331.3) = -1.85, p > .05 (β = -0.073, SE = .006, CI [-0.025 0.001]), he
g ow h be ween he i s wo s ages was e iden in Spanish, (331.4) = 4.89, p < .001 (β
= .031, SE = .006, CI [0.019 0.044]). Be ween he 2nd and 3 d s ages, he e was a
signi ican inc ease in bo h languages, bu such inc ease was s onge in Spanish,
(327.4) = 12.67, p < .001 (β = .083, SE = .007, CI [0.07 0.096]), han in Basque, (328) =
7.75, p < .001 (β = .073, SE = .009, CI [0.055 0.092]). Thus, he signi ican gap be ween
bo h languages (i.e., Basque > Spanish) p esen in s age 1, (326.6) = 11.68, p < .001 (β =
.073, SE = .006, CI [0.061 0.086]), closed g adually (s age 2: (326.6) = 4.57, p < .001 (β
= .03, SE = .007, CI [0.017.043])), and ecep i e ocabula y pe o mance was no
signi ican ly di e en be ween languages by s age 3, (326.1) = -0.66, p > .05 (β = -0.004,
SE = .007, CI [-0.017 0.009]).
33
The LME model o he hi d o ou p oxies o lexical abili ies, lexical di e si y,
yielded ha longi udinal s ages we e signi ican p edic o s o change F(2, 193.54) = 13.15,
p < .001, while language did no p edic signi ican di e ences, F(1, 224.71) = 0.97, p >
.05. The e was also a signi ican in e ac ion be ween language and s age, F(2, 194.9) =
7.33, p = .001, e lec ing di e en longi udinal ajec o ies o lexical di e si y measu es
o Basque and Spanish (Figu e 1.1). In he case o Basque, he e was no a signi ican
g ow h be ween he 1s and 2nd s ages, (182.2) = .62, p > .05 (β = .005, SE = .008, CI [-
0.011 .02]) whe eas be ween s ages 2 and 3, lexical di e si y inc eased, (179.5) = 6.11, p
< .001 (β = .05, SE = .008, CI [.033 .064]). In Spanish, he e was no signi ican g ow h
in lexical di e si y du ing he imespan o he longi udinal s udy (di e ence be ween
he 1s and 3 d s ages: (207.4) = 1.33, p > .05 (β = .016, SE = .012, CI [-0.023 .021])). Such
g ow h in Basque lexical di e si y be ween he 2nd and 3 d s ages, in con as o Spanish
non-signi ican g ow h, p oduced a signi ican be ween-languages di e ence in he 3 d
s age o he s udy. Thus, lexical di e si y was highe in Basque han in Spanish by he
las s age o he s udy, (212) = 3.73, p < .001 (β = .035, SE = .009, CI [.016 .053]).
Syn ac ic abili ies
Nex , we i ed one LME o each o ou h ee syn ac ic me ics. Wi hin
ecep i e syn ac ic span abili ies, we obse ed ha he pe o mance in sen ence
epe i ion was signi ican ly modula ed by language, F(1, 313.66) = 19.62, p < .001, and
s age, F(2, 320.09) = 151.54, p < .001, as well as by he in e ac ion o bo h ac o s, F(2,
312.14) = 20.46, p < .001. The in e ac ion be ween language and s age was d i en by he
ac ha Spanish sen ence epe i ion g ew a a as e a e han Basque du ing he
imespan o ou s udy (Figu e 1.1). Thus, while sen ence epe i ion pe o mance was no
34
signi ican ly di e en be ween Basque and Spanish in he 1s , (312.2) = .12, p > .05 (β =
.098, SE = .805, CI [-1.485 1.681]), no he 2nd s age, (312.9) = .05, p > .05 (β = .041, SE
= .846, CI [-1.624 1.705]), by he 3 d s age, sen ence epe i ion in Spanish was al eady
signi ican ly highe han in Basque, (312.8) = 7.64, p < .001 (β = 6.57, SE = .86, CI [4.88
8.266]).
Rega ding syn ac ic complexi y measu es, i is wo h no ing ha , in line wi h
p e ious s udies (Nippold, 2009; Nippold e al., 2005, 2007), we ound s ong
co ela ions be ween MLU and clausal densi y (Supplemen al Figu e 1.2). MLU-15 was
signi ican ly modula ed by language, F(1, 245.85) = 11.43, p = .001, and longi udinal
s age, F(2, 195.46) = 12.16, p < .001, and ha he e was no in e ac ion be ween bo h
ac o s, F(2, 193.78) = 2.4, p > .05. Thus, MLU-15 was o e all highe in Basque han in
Spanish, (245.8) = 3.38, p <.001 (β = 1.067, SE = .315, CI [.445 1.687]). O e all,
longi udinal g ow h in MLU-15 was only obse ed be ween he 2nd and 3 d s ages, (185)
= 3.9, p < .001 (β = 1.39, SE = .355, CI [.684 2.087]). None heless, Basque MLU-15 g ew
be ween he 1s and 2nd s ages, (178.9) = 2.17, p = .031 (β = .898, SE = .414, CI [.081 1.715]),
while Spanish MLU-15 did no show signi ican g ow h in he same ime window,
(206.7) = 1, p > .05 (β = .246, SE = .634, CI [-1.496 1.005]), as isible in Figu e 1.1.
The LME model o clausal densi y a signi ican e ec o s age, F(2, 157.76) =
4.47, p = .013, no signi ican o e all di e ence be ween languages, F(1, 169.9) = 2.19, p >
.05, and a ma ginal in e ac ion be ween language and s age, F(2, 160.96) = 3.04, p = .05.
Since his in e ac ion was ma ginal, i was u he explo ed. Unlike in MLU-15, CD
g ow h in Basque did no ake place be ween he 1s and 2nd s ages, (150.3) = 1.88, p >
.05 (β = .033, SE = .018, CI [-0.002 .069]). Howe e , he e was a signi ican inc ease

35
be ween he 2nd and 3 d s ages, (146) = 2.69, p = .008 (β = .047, SE = .018, CI [.013 .082]),
such longi udinal g ow h was no s a is ically signi ican in Spanish (di e ence be ween
he 1s and 3 d s ages: (164.9) = .2, p > .05 (β = .006, SE = .03, CI [-0.053 .065])). Such
di e ing g ow h pa e ns be ween Basque and Spanish CD a e isible in igu e 1.1.
Phonological abili ies
We had o disca d hyme de ec ion om he analyses because i u ned ou o
be a oo di icul ask o ou pa icipan s, as he as majo i y did no unde s and he
ins uc ions (i.e., he hyme concep ) o pe o med below chance le el (50% o co ec
esponses). Rega ding he emaining nonwo d epe i ion ask, LME model yielded
signi ican main e ec s o language, F(1, 325.4) = 28.11, p < .001, s age, F(2, 330.36) =
48.65, p < .001, and no signi ican in e ac ion be ween language and s age, F(2, 325.44)
= 1.63, p > .05. Such longi udinal pa e n, ep esen a i e o an o e all g ow h in nonwo d
epe i ion ha was pa allel be ween Basque and Spanish, is isible in he Figu e 1.1.
Thus, nonwo d epe i ion pe o mance in Basque was signi ican ly highe in Basque
han in Spanish ac oss he s udy, (325.4) = 5.3, p < .001 (β = .053, SE = .010, CI [.033
.073]), and he e we e signi ican inc eases in bo h languages be ween he 1s and 2nd
s ages, (332.1) = 2.94, p = .004 (β = .036, SE = .012, CI [.012 .06]), and he 2nd and 3 d
s ages espec i ely, (326.1) = 6.69, p < .001 (β = .084, SE = .012, CI [.059 .108]).
36
Figu e 1.1. Longi udinal ajec o y o he di e en language measu es o he s udy, g ouped by
lexico-seman ic (g een), syn ac ic (pu ple), phonological (o ange), and AoE measu es (da k
g ey). Poin s ep esen each pa icipan ’s sco e in each measu e, language, and s age. Boxplo s
ep esen g oup es ima es, wi h ho izon al lines wi hin each box ma king he median sco e.
Uppe and lowe hinges ma k he i s and hi d qua ile, and whiske s show 1.5 * in e -qua ile
ange. The connec ing lines ep esen longi udinal ajec o ies be ween wo consecu i e s ages:
con inuous and do ed lines ma k signi ican and non-signi ican g ow h, espec i ely.
Signi ican be ween-languages di e ences wi hin each s age a e ma ked wi h *, **, and *** o
p < .05, 01, and .001 espec i ely. The symbol · ma ks ma ginally signi ican di e ences (i.e., p
= .05).
O e all, he e was g ow h ac oss he di e en language domains du ing he
imespan o ou longi udinal s udy in Basque and Spanish. None heless, he e we e
37
di e en ial a es o g ow h be ween languages o mos o he linguis ic domains, which
could be a ibu ed o he ma ked di e ences in AoE ha a e p esen in ou sample.
The e o e, he aim o ou second se o analyses was o assess whe he sus ained
languages di e ences in AoE ha e a di ec in luence on language pe o mance ac oss
he 3 longi udinal s ages was he speci ic aim o ou second se o analyses. The lack o
signi ican changes in AoE ac oss he longi udinal s ages (see las panel o Figu e 1.1)
was an impo an assump ion o each his objec i e, assump ion which was espec ed
by ou da a.
1.3.2. Di ec In luence o amoun o exposu e o Basque and Spanish ac oss
domains
1.3.2.1. Fi s subse o models: The e ec o AoE on each measu e ac oss languages
Lexico-seman ic abili ies
AoE was posi i ely ela ed o p oduc i e ocabula y in bo h languages
h oughou he whole longi udinal s udy, (327.7) = 13.07, p < .001 (β = .039, SE = .003).
I s in luence on pic u e-naming ask was no signi ican ly di e en be ween languages,
(95.6) = 1.01, p > .05 (β = .007, SE = .007).
Simila ly, ecep i e ocabula y was g ea ly modula ed by AoE in bo h
languages, (381) = 7.17, p < .001 (β = .02, SE = .003), wi h no signi ican be ween-
languages di e ence in he in luence o AoE on wo d comp ehension, (381) = 1.65, p >
.05 (β = .009, SE = .006).
38
Fo he lexical di e si y indexes, we did no obse e an o e all in luence o AoE,
(228) = 78, p > .05 (β = .003, SE = .004), no a di e en in luence o AoE be ween
Basque and Spanish, (382) = -1.9, p > .05 (β = -0.012, SE = .006).
Syn ac ic abili ies
O e all, AoE was a ele an p edic o o ecep i e syn ac ic span, (310.94) =
9.11, p < .001 (β = 2.971, SE = .326), wi h no signi ican di e ence in i s in luence be ween
Basque and Spanish on he epe i ion o sen ences, (98.19) = 1.27, p > .05 (β = 1.12, SE =
.882).
MLU-15 measu ed du ing spon aneous speech p oduc ion was signi ican ly
p edic ed by AoE in bo h languages, (228.98) = 6.3, p < .001 (β = 1.119, SE = .177), wi h
no signi ican in e ac ion o AoE and language, (85.06) = -0.96, p > .05 (β = -0.318, SE
= .332). Howe e , AoE was no an o e all signi ican p edic o o he clausal densi y o
he spon aneous speech p oduc ions o child en, (146.69) = .47, p > .05 (β = .004, SE =
.01), and he e was no signi ican di e ence in he in luence o AoE on his measu e
be ween languages, (91.05) = -0.75, p > .05 (β = -0.012, SE = .016).
Phonological abili ies
AoE was no di ec ly ela ed o phonological sho - e m memo y abili ies as
i did no p edic nonwo d epe i ion pe o mance signi ican ly, (146.69) = .47, p > .05
(β = .004, SE = .01). In addi ion, he e was no signi ican di e ence o he in luence o
AoE on his measu e be ween languages, (91.05) = -0.75, p > .05 (β = -0.012, SE = .016).
45
The ac ha he g ow h o p oduc i e lexical di e si y and syn ac ic complexi y
and hei link wi h AoE mainly ake place a he las de elopmen al s age o ou s udy
(a ound 6 yea s o age) con as s wi h he sus ained g ow h and in luence o AoE on
ecep i e and p oduc i e ocabula y knowledge, and ecep i e syn ac ic span. Exposu e
migh he e o e play a ele an ole in explaining language pe o mance bu his may
a y depending on he s age o de elopmen and he language domain and abili y
conside ed. A he beginning o he s udy, child en elied on AoE o build hei lexical
ecep i e knowledge. In pa allel wi h he de elopmen al inc ease in ocabula y
knowledge, we obse ed ha mo e linguis ic domains s a ed o depend on exposu e.
This sugges s ha ocabula y knowledge migh ac as a building block o he
de elopmen o b oade a ays o language abili ies (see also Simon-Ce eijido &
Gu ié ez-Clellen, 2009). In e es ingly, a s eadily di e se pa en al ocabula y inpu was
shown o de e mine ocabula y knowledge o monolingual child en ea ly on ( om 1 o
2.5 yea s o age; Sil ey e al., 2021), in line wi h ou inding ha he s onges in luence
o AoE on ecep i e and p oduc i e ocabula y ac oss languages was a he ea lies poin
o ou s udy (4 yea s o age). Howe e , syn ac ic complexi y measu es migh
inc easingly bene i om exposu e h oughou de elopmen ( able 1.2) (see also Sil ey
e al., 2021).
Simila o p oduc i e ocabula y, he e was a s eady and pa allel g ow h o
phonological abili ies ac oss languages. Mos in e es ingly, nonwo d epe i ion was no
in luenced by indi idual AoE indexes in nei he language no de elopmen al s age (see
able 1.2). The limi ed ole o AoE on phonological compa ed o lexical and syn ac ic
de elopmen (C is ia, 2020) migh s em om he s onge dependence o phonological
acquisi ion on he i able gene ic ac o s compa ed o en i onmen al ac o s such as AoE

46
(Bishop, 2002; Ko as e al., 2005). None heless, phonological pe o mance was o e all
be e when child en had o epea nonwo ds wi h syllabic and s ess phono ac ics
speci ic o hei dominan language also associa ed wi h highe le els o exposu e,
despi e he highly simila phonological Basque and Spanish sys ems (Ezeizaba ena &
Ga cía, 2015). Al oge he hese indings suppo a ela i ely small and indi ec ole o
AoE in he de elopmen o language-speci ic phono ac ics sensi i i y (Messe e al.,
2010; Munson, 2001; Vi e i ch & Luce, 2005) compa ed o o he ac o s (e.g., language
p o iciency lexical knowledge, Gu ié ez-Clellen & Simon-Ce eijido, 2010). I is
no ewo hy ha because he nonwo ds o he epe i ion ask ollowed language-speci ic
lexical s ess pa e ns, a ia ions in lexical knowledge be ween languages migh ha e
di ec ly con ibu ed o pe o mance on ou phonological ask.
O e all, he p esen s udy in o ms heo e ical and empi ical esea ch ha
p oposes ha he ole o inpu in language acquisi ion is modula ed by language
domains and de elopmen al s ages (e.g., Chond ogianni & Ma inis, 2011).
Mo e pa icula ly, i s , ocabula y lea ning appea s as a ‘da a-hung y’ p ocess
ha bene i s g ea ly om he a ailable inpu e en when i is limi ed (C is ia, 2020):
ocabula y knowledge inc eased in bo h languages h oughou he s udy and was
g ea ly in luenced by AoE. In addi ion, he ela i ely la e g ow h o p oduc i e lexical
di e si y in he mo e dominan language o bilingual child en seemed o di ec ly
depend on AoE, possibly e lec ing a as e ocabula y de elopmen in his language.
Second, we obse ed a obus g ow h in he p oduc i e syn ac ic complexi y o
child en in bo h languages. This syn ac ic g ow h depended igh ly on child-speci ic
AoE indexes as i occu ed when AoE was a signi ican p edic o o hese abili ies. We
47
sugges ha la e s ages o language de elopmen (as compa ed o ocabula y) could be
especially sensi i e o and de e minan o syn ac ic de elopmen .
Las , we showed ha he de elopmen o phonological abili ies s eadily ook
place in bo h languages and ha AoE did no di ec ly in luence such acquisi ion as i
was he case o he o he language domains assessed. None heless, i canno be uled
ou ha AoE migh in luence he sensi i i y o language-speci ic phono ac ics ha we
know play a ole in building-up phonological ep esen a ions.
1.4.1. Conclusion
In summa y, ou indings highligh a de elopmen al igh ela ionship be ween
AoE and bilingual language de elopmen a di e en le els o linguis ic g anula i y and
domains ( om sublexical phonological s uc u e o lexical and syn ac ic cons uc ions).
Ou esul s suppo he usage-based heo ies o language acquisi ion (Tomasello, 2005)
and help ex end such amewo ks o bilingual language acquisi ion. Ou indings
con ibu e o he ield by shedding new ligh on how child-speci ic language
en i onmen s shape bilingual language de elopmen al ajec o ies and he language
“gap” ha cha ac e izes he pe o mance o unbalanced bilinguals. O e all, we o e a
de elopmen al snapsho o he in e play be ween child en in insic cha ac e is ics and
en i onmen al ac o s on he eme gence o p o icien language knowledge and use,
pa ing he way o he design o mo e ailo ed and e icien p og ams o bilingual
educa ion and he assessmen o language diso de s in his g oup o child en.
48
S udy 1 a es ed he ele an in luence o AoE on he de elopmen o
undamen al language domains in wo languages o bilingual child en. While S udy 1
ocused on he de elopmen al ajec o y o di e en language domains om a
beha io al pe spec i e, he goal o S udy 2 was o shed ligh on he ole o AoE on he
neu ocogni i e mechanisms which suppo language comp ehension, such as he
co ical acking o speech.
49
2. S udy 2. The con ibu ion o ea ly language
exposu e o he co ical acking o speech: e idence
om bilingual child en
2.1. In oduc ion
Con inuous exposu e o spoken language in a wide a ie y o con ex s makes i s
acquisi ion appea spon aneous and e o less. None heless, highly complex b ain
mechanisms sub end he p ocesses ha allow us o unde s and and speak a language
luen ly. Language comp ehension equi es e icien analysis, segmen a ion and pa sing
o he apidly un olding linguis ic s uc u es ha a e embedded in con inuous speech
s eams. De elopmen al e idence shows ha he co ical mechanisms suppo ing
language comp ehension eme ge and become language-speci ic essen ially be ween
bi h and he s a o p ima y school ( o e iews see Kuhl, 2004; and Skeide &
F iede ici, 2016). Howe e , li le is known abou whe he and how hese
neu ophysiological p ocesses ma u e as a unc ion o he amoun o exposu e (AoE) o
a language du ing ea ly childhood, despi e AoE being a s ong con ibu o o he
de elopmen o language knowledge in hese ea ly s ages (S udy 1). In he p esen s udy,
we in es iga ed whe he AoE and language knowledge in luenced neu al oscilla o y
ac i i y in esponse o speech in bilingual child en. Limi a ions in AoE could ha e
downs eam consequences o he de elopmen o such neu ocogni i e abili ies du ing
ea ly childhood and a ec child en’s u u e social and academic wellbeing. Addi ionally,
he numbe o child en who a e exposed o se e al languages om an ea ly age and
he eby ecei e limi ed inpu s in each o hei languages, is apidly inc easing (Pa adis
e al., 2011). Thus, accu a e mul idimensional models abou he de elopmen o
50
language skills equi e unde s anding how di e en le els o language exposu e and
p o iciency modula e undamen al b ain mechanisms unde lying speech
comp ehension.
We hypo hesized ha AoE and language p o iciency shape he ma u a ion o he
co ical acking o speech (CTS) —a neu ocogni i e p ocess ha has been shown c i ical
o unde s anding con inuous speech— o phonological, lexical and syn ac ic
in o ma ion. We p opose ha AoE is de e mining in building up speech empo al
s a is ics ha con ibu e o he e icien dynamic alignmen o co ical oscilla o y
ac i i y o ele an uni s in he speech signal ( he so-called speech-b ain en ainmen ;
Gi aud & Poeppel, 2012; Luo & Poeppel, 2007). Thus, g ea e AoE o a gi en language
should be associa ed wi h mo e p ecise CTS. Addi ionally, iche language knowledge
p o ided by AoE should suppo he acking o linguis ic in o ma ion a di e en
linguis ic le els (i.e., phonological, lexical and syn ac ic) o help he e icien and expe
comp ehension o con inuous speech (B ode ick e al., 2021; en Oe e & Ma in, 2021).
S udy 1 showed a clea con ibu ion o AoE o language knowledge, which also
depended on he language domain and he de elopmen al s age conside ed (see also V.
C. M. Ga he cole & Thomas, 2005; Olle & Eile s, 2002; Pa adis & Jia, 2016). In
monolingual en i onmen s, he amoun o child-di ec ed inpu has been ela ed o
wo d p ocessing speed (Hu ado e al., 2008; Weislede & Fe nald, 2013) ha is a
ele an index o he encoding o lexical in o ma ion om a con inuous speech inpu .
Hu ado e al. (2014) ound ha , in bilingual en i onmen s, he ela i e AoE o wo
languages was posi i ely co ela ed wi h wo d p ocessing speed in each language
sepa a ely. The e o e, AoE appea s o p o ide child en wi h language knowledge a

51
di e en le els ha , in u n, suppo s inc easingly e icien p ocessing and
comp ehension skills o con inuous speech.
Impo an ly, exposu e enhances ce ain aspec s o phonological abili ies
(Ni oue , 1996; Ni oue & Bu on, 2005; o a e iew, see Ni oue , 2002), which
ha e been p oposed o be sub ended by CTS (Goswami, 2011, 2017; as de ailed below).
Mo e speci ically, da a has shown ha he in luence o AoE on language lea ning du ing
childhood is media ed by he p og essi e uning o phonological abili ies, ha acili a e
he comp ehension o speech as i un olds o e ime (S. E. Ga he cole, 2006; S. E.
Ga he cole e al., 1991; Pa a e al., 2011). Al hough in S udy 1 we did no ind a di ec
ela ionship be ween AoE and phonological abili ies, we did obse e ha he
phonological abili ies in he mo e dominan language de eloped ea lie han in he less
dominan language. Thus, i appea s ha su icien language knowledge is equi ed o
phonological ep esen a ions o eme ge and gain language speci ici y. Fo example, AoE
o he di e en languages o bilinguals bene i s hei pe o mance when epea ing
language-speci ic nonwo ds (Messe e al., 2010; Pa a e al., 2011). In addi ion, esea ch
in child en wi h hea ing di icul ies show ha educ ions in he quali y and quan i y o
speech inpu delay he de elopmen o bo h phonological and b oade speech
comp ehension skills (B iscoe e al., 2001; Ni oue & Bu on, 2001, 2005). O e all, he
a o emen ioned s udies s ongly sugges ha an e icien CTS de e mines adequa e
phonological de elopmen .
Goswami (2011, 2017) p oposed ha he co ical acking o ampli ude
modula ions (AMs) — empo al luc ua ions in he speech signal— a del a (0.5 – 4 Hz)
and he a (4 – 8 Hz) equency bands, aligning closely wi h he occu ence o
52
s ess/p osodic pa e ns and syllables espec i ely, subse es he empo al sampling o
ph ases, wo ds, and syllables du ing phonological de elopmen . Acco dingly, he
accu a e p ocessing o AMs con ibu es o he eme gence o phonological
ep esen a ions (Goswami & Leong, 2013; Leong & Goswami, 2014). In adul s, CTS a
he del a and he a equency bands has been shown o suppo speech comp ehension
(e.g., Ding & Simon, 2012; G oss e al., 2013; Luo & Poeppel, 2007; Molina o & Liza azu,
2018; Peelle e al., 2013), and a g owing numbe o s udies ha e shown ha CTS is in
place and de eloping in in ancy and childhood (e.g., A ahe i e al., 2022; K. H. Menn,
Wa d, e al., 2022; Ríos-López e al., 2020).
The con ibu ion o inpu o phonological de elopmen h ough CTS is hin ed
by se e al pieces o e idence. Kalashniko a e al. (2020) showed ha , when add essing
child en a isk o de elopmen al dyslexia (cha ac e ized ypically by phonological
de ici s), adul s’ use o hype a icula ed owels was educed, which could ha e an
impac when accessing phonological in o ma ion om he speech signal. Also ele an
in his ega d, se e al s udies showed ha phonological de ici s in dyslexic child en a e
igh ly linked o a ypical CTS wi hin he del a and he a bands (Des oky e al., 2020; Di
Libe o e al., 2018; G anados Ba be o e al., 2022; Molina o e al., 2016; Powe e al.,
2016) in compa ison o bo h ch onological-age-ma ched and eading-age-ma ched
pee s (Di Libe o e al., 2018; Powe e al., 2016), which sugges s ha CTS could be
causally ela ed o phonological de elopmen . Howe e , Des oky e al. (2020) epo ed
poo e CTS in dyslexic child en only when compa ed o ch onological and no o
eading age-ma ched con ols. Gi en hese somewha di e gen indings, he
con ibu ion o accumula ed linguis ic expe ience o he ma u a ion o CTS du ing
phonological de elopmen emains an unanswe ed ques ion ha dese es explo a ion.
53
Ou s udy enables add essing whe he linguis ic inpu con ibu es o phonological CTS,
by exploi ing he non- i ially di e en AoEs o unbalanced bilinguals and es ing
concu en ly CTS and phonological abili ies in each o hei languages.
In addi ion o phonology, he e is also e idence sugges ing ha lexical and
syn ac ic knowledge (whose acquisi ion is igh ly linked o AoE, as we showed in S udy
1) migh be also linked o he e iciency o speech acking neu ocogni i e mechanisms.
CTS has been shown o be modula ed by knowledge abou bo h con ex -d i en wo d
p edic abili y (B ode ick e al., 2021; Klimo ich-G ay e al., 2021; Koskinen e al., 2020;
Molina o e al., 2021) and speech syn ac ic s uc u es (Kau eld e al., 2020; Meye e al.,
2017; Meye & Gumbe , 2018). In addi ion, Panda e al. (2020) showed ha ocabula y
knowledge was linked o he synch oniza ion be ween language- ela ed co ical a eas
h ough neu al oscilla o y ac i i y – al hough no speci ically CTS - du ing con inuous
speech lis ening.
Taken oge he , he a o emen ioned e idence poin s o he ole o acqui ed
language knowledge ac oss a ious linguis ic domains on CTS uning. Howe e , no
s udy di ec ly add essed he impac o language knowledge on CTS du ing ea ly
language lea ning, no whe he AoE could shape CTS. Shedding ligh on hese ques ions
could help unde s and he ole o CTS du ing de elopmen , which o da e, has been
essen ially ocused on he acquisi ion o phonology, and no on b oade non-
phonological language skills ha a e s ongly in luenced by AoE.
The aim o he p esen s udy was o explo e he ole o AoE and language
p o iciency on he co ical acking o phonological and non-phonological speech
ea u es o answe he ollowing ques ions:
54
1) Is he e a di ec in luence o AoE on he co ical acking o speech
dis ibu ional p ope ies coding o (i) phonological (i.e., he speech
en elope) and (ii) non-phonological (i.e., lexical and syn ac ic s uc u es)
in o ma ion?
2) Does he co ical acking o speech ea u es coding o phonological, lexico-
seman ic and syn ac ic in o ma ion p edic beha io al language pe o mance
in hese linguis ic domains espec i ely?
2.2. Me hod
2.2.1. Pa icipan s
Pa icipan s we e 35 (18 emales) Basque-Spanish bilinguals be ween 6 and 7
yea s o age (Mean age = 6.92; SD = .11). They we e he child en om S udy 1 who showed
he mos exposu e o Basque (> 70% o hei waking hou s) and he leas o Spanish (<
30%) when hey we e 6 yea s old (a he las s age o S udy 1). Such selec ion c i e ia
we e aimed o enable us o in es iga e he in luence o AoE (low in Spanish s. high in
Basque) on CTS in wo di e en languages wi hin he same pa icipan s.
Since he pa icipan s had s a ed o mal eading ins uc ion only a ew mon hs
be o e he s a o he s udy, he amoun o w i en inpu was no su icien o be
conside ed a main sou ce o language inpu a he ime o es ing. Thus, he po en ial
in luence o w i en language exposu e on he con ibu ion o AoE o CTS in ou s udy
was conside ed o be minimal (Dehaene e al., 2015; Goswami & B yan , 1990; Mo ais e
al., 1979).
61
senso s was only used o de e mine he ime o maximum cohe ence, and no o
localizing cohe ence e ec s in ou analyses (which we e loca ed h ough clus e -based
pe mu a ion es s, see la e ). A 60 ms lag o he EEG wi h espec o he speech signal
was he imepoin o maximum cohe ence ac oss pa icipan s and condi ions, o which
we ci cumsc ibed ou speech-b ain cohe ence es ima es o la e s a is ical analyses.
mTRFs
Th ee mTRF models we e compu ed, coding o dis ibu ional ea u es o
speech a h ee le els: en elope (phonology), lexical equency (lexical), and sen ence-
le el seman ic dis ance (syn ac ic). Impo an ly, compu ing en elope mTRF was used
o e i y ha i ela ed o he en elope acking measu ed wi h cohe ence and o
es ablish a posi i e con ol o mTRF analyses o non-phonological ea u es, which a e
less salien in he signal.
The speech en elope eg esso s ha e he same ope a ionaliza ion (i.e., he
Hilbe ans o med audio signal) as desc ibed in he speech-b ain cohe ence analyses
(also ollowing he speci ica ions in Molina o & Liza azu, 2018).
Lexical and syn ac ic eg esso s consis ed o con inuous ec o s (one pe
linguis ic ea u e and s o y) consis ing o bu s s a he onse o e e y con en wo d in
each s o y. The ampli ude o such bu s s co esponded o la en a iables ha we e used
as p oxy o lexical and syn ac ic in o ma ion espec i ely. Fo he lexical eg esso , we
used lexical equency: he ampli ude o each bu s was he in e se o he Zip lexical
equency o i s co esponding wo d (simila o he ope a ionaliza ion o Zip equency
in S udy 1; see Supplemen al Figu e 2.1). Fo he syn ac ic eg esso , we compu ed he
sen ence-le el seman ic dis ance ec o (see Supplemen al Figu e 2.2): we i s ob ained

62
he seman ic ep esen a ion o each s o y wo d in a 300-dimensional space h ough
as ex Py hon package o ex ep esen a ions (Joulin e al., 2016) and hen, he
ampli ude o each bu s was compu ed as 1 minus he Pea son co ela ion o he
seman ic dimensions o a wo d wi h he a e age o all i s p eceding wo ds wi hin a
sen ence. This way, lexical i ems wi h a bigge seman ic co ela ion wi h hei p eceding
wo ds wi hin a sen ence we e less salien , and less seman ically ela ed wo ds s ood ou
ela i e o hei p eceding con ex .
A e ob aining all he eg esso s o in e es , we used he mTRF oolbox (C osse
e al., 2016) in MATLAB o i one mTRF encoding model wi h he EEG esponse o
each ea u e (en elope, lexical equency, and sen ence-le el seman ic dis ance), and
o each language and pa icipan . Ou mTRF analysis ime window was 900 ms-long,
spanning om 150 ms be o e o 750 ms a e e e y alue in each eg esso . In o de o
ain and es he encoding model, we spli ou con inuous EEG signal and each
co esponding ea u e ec o s (~ 16 minu es) in o 8 olds o equal leng h (~ 2 minu es).
We ained he encoding model in 7 olds and es ed i s accu acy in he emaining one,
epea ing his p ocess o 30 i e a ions pe old. In each i e a ion, we ob ained a
co ela ion coe icien ( - alue) be ween each ea u e and he EEG esponse as well as
be ween he EEG esponse and a andomly pe mu ed e sion o he ea u e. Also, wi hin
each i e a ion, he co ela ion coe icien s o each ea u e we e con as ed agains hei
pe mu ed e sion, and he esul s we e a e aged ac oss i e a ions. The esul ing a e age
alue, he co ela ion coe icien be ween a ea u e and EEG abo e chance le el, was
used in u he s a is ical analyses as p oxy o he ex en o which each ea u e linea
mapped by EEG ac i i y.
63
Figu e 2.1. G aphical summa y o he CTS analyses. In blue, he speech wa e o m o “En el
uni e so, hay cien os de miles de millones de galaxias” (“In he uni e se, he e a e hund eds o
billions o galaxies”). Speech-b ain cohe ence and en elope-le el mTRF models a e based on he
ela ionship be ween he speech en elope (in ed) and EEG ac i i y ( op). Lexical equency and
sen ence-le el seman ic dis ance mTRFs a e ob ained om he EEG esponse o bu s s o
di e en ampli ude (lexical equency, o ange; sen ence-le el seman ic dis ance, pink) a he
onse o each con en wo d.
2.2.4. S a is ical analysis
2.2.4.1. Beha io al language measu es
In o de o assess be ween-languages di e ences in language pe o mance, we
i ed LME models wi h language as p edic o o each dependen measu e (AoE and
sco es on he pic u e naming, nonwo d epe i ion, and nonwo d eading asks) and
64
pa icipan s as andom in e cep s, o accoun o he wi hin-indi idual design o ou
s udy. LME models we e i ed using he same so wa e speci ica ions han in S udy 1.
2.2.4.2. Speech-b ain cohe ence
We used clus e -based pe mu a ion es s (CBPTs) o analyze whe he he e we e
signi ican speech-b ain cohe ence e ec s in Basque and Spanish, as well as be ween-
languages di e ences in such CTS me ic. CBPT is an e icien way o es ima ing he
p esence o a s a is ical e ec in a high dimensional space, as i allows o accoun o
he spa ial adjacency o elec odes and es o signi ican e ec s ha a e sha ed ac oss
a g oup o elec odes (a clus e ) (Ma is & Oos en eld, 2007). In ou case, we un
dependen -samples one- ailed CBPTs in FieldT ip wi h 1000 pe mu a ions. CBPTs
allowed o a minimum o 2 elec odes as clus e as a way limi he possibili y o single-
elec ode alse-ala m e ec s, and we co ec ed o mul iple compa isons based on he
numbe o a-p io i signi ican clus e s. We i s assessed whe he he e was abo e-
chance speech-b ain cohe ence by con as ing h ough CBPT he phase alignmen o
EEG and genuine speech en elope e sus he phase alignmen o EEG and a su oga e
e sion o he en elope ha did no ollow he o iginal speech o de (i.e., lipped speech
en elope su oga e). Then, we con as ed (also h ough CBPT) he speech-b ain
cohe ence alues o Basque and Spanish o see whe he he e was a signi ican language
e ec in his CTS me ic.
2.2.4.2. mTRFs
Fo be ween-languages compa isons o mTRFs, we used each pa icipan ’s
co ela ion coe icien ( - alue) be ween he eg esso (e.g., lexical equency) and he
65
EEG esponse as indi idual es ima e o how ai h ully EEG mapped a gi en speech
ea u e (e.g., lexical equency). Thus, an - alue o 1 would be a pe ec co espondence
be ween mTRF ea u e and EEG signal, and an - alue o 0 would mean no co ela ion
wha soe e . We selec ed he a e age - alue o he 5 elec odes ha showed he bigges
co ela ion wi h each eg esso o s a is ical compa isons be ween languages. Speech-
b ain cohe ence yields es ima es a senso le el, which can be submi ed o CBPT o
es ima e he opog aphical loca ion o he e ec on he scalp. Gi en ha ou es ima e
o he mTRF model i was a single - alue pe pa icipan , language and ea u e, we did
no ha e he senso -le el esolu ion necessa y o CBPT. Thus, ins ead o CBPTs, we
used Bayesian - es s (in JASP, e sion 0.16.4, JASP eam, 2022) o con as whe he he
mTRF models o en elope, lexical, and syn ac ic in o ma ion di e ed o no be ween
Basque and Spanish. Bayesian - es s allow o assess no only whe he he e was
e idence o he al e na i e hypo hesis (in ou case, be ween-languages di e ence in
CTS), bu also o he null hypo hesis (be ween-languages simila i y in CTS) ( an Doo n
e al., 2021).
2.2.4.3. Co ela ion analyses be ween CTS and beha io al language measu es
In o de o explo e ela ionships be ween CTS o di e en linguis ic ea u es and
language pe o mance, linea eg essions we e conduc ed wi hin each language, by
co ec ing o amily-wise ype I e o ia alse disco e y a e (FDR).
66
2.3. Resul s
2.3.1. Language measu es
In line wi h hei signi ican ly highe exposu e o Basque han o Spanish, (63)
= 35.13, p < .001 (β = 72.298, SE = 2.058), pa icipan s showed signi ican ly highe
ocabula y knowledge, (32) = 13.18, p < .001 (β = 0.077, SE = .006), and nonwo d
eading decoding, (30.35) = 5.02, p < .001 (β = 3.263, SE = .651), in Basque han in
Spanish (see Figu e 2.2). Howe e , he e was no language e ec on nonwo d epe i ion,
(30.1) = -0.84, p > .05 (β = -0.014, SE = .017), no on he comp ehension o he s o ies
ha pa icipan s lis ened o du ing he EEG session, (28) = 0.12, p > .05, (β = 0.005, SE
= .039).

67
Figu e 2.2. Amoun o exposu e and beha io al pe o mance in ocabula y, phonology, and
eading as a unc ion o es ing s age. Poin s ep esen each pa icipan ’s sco e in he di e en
measu es, languages, and s ages. Boxplo s ep esen g oup es ima es, wi h ho izon al lines
wi hin each box ma king he median sco e. Uppe and lowe hinges ma k he i s and hi d
qua ile, and whiske s show 1.5 * in e -qua ile ange. Lines connec he sco es o each
pa icipan be ween languages.
2.3.2. En elope CTS
Wi hin he del a equency band, speech-b ain cohe ence was signi ican
be ween 0.5 and 1.5 Hz in Basque, clus e s a is ic = 271.73, p < .001, (SD = 0.001), and
Spanish, clus e s a . = 278.17, p = .001, (SD = 0.001) (Figu e 2.3). In he he a ange (4
– 7 Hz), we did no ind signi ican speech-b ain cohe ence in any o he languages (all
co ec ed p- alues > .05). We also did no obse e signi ican cohe ence in he speci ic
5.5 Hz bin ha aligned closely o he syllable a e o bo h languages (p > .05).
In addi ion, he e was no language e ec on cohe ence in he men ioned 0.5 –
1.5 Hz del a band, in which bo h languages showed signi ican CTS (see also Figu e 2.3).
Mo eo e , cohe ence in his del a ange had conside ably o e lapping opog aphies in
bo h languages (Figu e 2.4).
68
Figu e 2.3. Speech-b ain cohe ence ac oss equency bands o Basque ( ed), Spanish (blue)
agains hei andom su oga e ( lipped e sion); and he con as be ween Basque and Spanish
(g een). The discon inuous ho izon al lines on he bo om ma k he equency ange in which
he e was signi ican cohe ence o Basque ( ed), and Spanish (blue).
69
Figu e 2.4. Topog aphy o speech-b ain cohe ence in he 3 signi ican equencies in Basque
( op) and Spanish (bo om). The colo map ma ks he size o he di e ence in cohe ence
(no malized) be ween genuine speech and i s lipped e sion (i.e., yellow, highe ela i e
cohe ence; blue, lowe ). Bigge do s wi hin each opog aphic map signal signi ican elec odes
(in CBPT).
Simila o speech-b ain cohe ence, we did no ind signi ican di e ences in
speech en elope mTRFs be ween Basque and Spanish. Indeed, he e was weak e idence
in a o o he lack o be ween-languages di e ences in he EEG esponses o speech
AMs, BF10 = .357, e o = .034 %, median di e ence = -0.189, CI [-0.522 .138]
(supplemen al Figu e 2.4).
2.3.3. Lexico-syn ac ic CTS: lexical equency and sen ence-le el seman ic
dis ance
The e was no language e ec on lexical equency mTRFs (Figu e 2.5). In his
case, he e was mode a e e idence o he lack o be ween-languages di e ences, BF10 =
70
.204, e o = .037 %, median di e ence = -0.065, CI [-0.369 .264] (Figu e 2.5, and
supplemen al Figu e 2.5).
In addi ion, we ound mode a e e idence agains a language di e ence on
sen ence-le el seman ic dis ance mTRF indexes, BF10 = .194, e o = .037 %, median
di e ence = -0.040, CI [-0.370 .289] (Figu e 2.5, and supplemen al Figu e 2.6).
Figu e 2.5. Language con as s o he co ela ions be ween eg esso s and EEG signal o mTRFs
o speech en elope, lexical Zip equency, and sen ence-le el seman ic dis ances. Poin s
ep esen each pa icipan ’s - alue in he di e en mTRF models. Boxplo s ep esen g oup
es ima es, wi h ho izon al lines wi hin each box ma king he median sco e. Uppe and lowe
hinges ma k he i s and hi d qua ile, and whiske s show 1.5 * in e -qua ile ange. Lines
connec each indi idual’s - alues.
2.3.4. Rela ionship be ween CTS and beha io al language measu es
An en elope CTS and a lexico-syn ac ic CTS composi e sco es we e compu ed o
agg ega e EEG me ics o a mo e obus es ima e o CTS-beha io ela ionships, while
limi ing he possibili y o spu iously signi ican indings. These composi e sco es we e
alida ed by he posi i e co ela ions ound be ween (i) cohe ence o he speech
en elope a del a and he a, and he en elope mTRF (a e age Pea son’s coe icien =
77
phonological skills (Leong & Goswami, 2014, 2017; Molina o e al., 2016), s udies in
adul s highligh ha CTS migh be mo e elian on speech a e a he han on he
speech en elope modula ions, a leas in quie lis ening en i onmen s (Hauswald e al.,
2022; Schmid e al., 2021). The e o e, he ole o en elope CTS o ea ly language
de elopmen migh dec ease as soon as indi iduals can ely on su icien lexico-
seman ic and syn ac ic knowledge o ack con inuous speech (e.g., Kau eld e al., 2020;
Molina o e al., 2021).
While he ole o language knowledge on he co ical acking o speech ea u es
beyond he pu e acous ic en elope is being inc easingly s udied in adul s (e.g.,
B ode ick e al., 2021; Kau eld e al., 2020), we epo ini ial e idence in child en. We
obse ed a signi ican posi i e ela ionship be ween lexical abili ies and he co ical
acking o lexico-syn ac ic ea u es con eyed by he speech signal, ci cumsc ibed o he
mos dominan language o he bilingual child en. These indings a e in line wi h
Hu ado e al. (2014)’s esul s showing ha , in in an s g owing in bilingual
en i onmen s, he ela i e AoE o hei languages was posi i ely co ela ed wi h speech
p ocessing e iciency and ocabula y knowledge in each language. Ou esul s add o
his pu ely beha io al e idence showing ha lexical knowledge abili ies a e ied o he
p ecise empo al mapping o lexico-syn ac ic con eyed in connec ed speech by
con inuous EEG ac i i y, bu only when sus ainedly high AoE is p o ided.
CTS p edic ed bo h phonological and lexical pe o mance only in Basque, he L1
o pa icipan s wi h he highes AoE and p o iciency (Figu e 2.6). We in e p e his as
e lec ing ha he mo e s able language ep esen a ions and p ocesses in Basque migh
mo e consis en ly and ai h ully map on o indi idual di e ences in CTS. On he

78
con a y, weake L2 Spanish knowledge migh no map as eadily on o neu al esou ces
engaged in CTS. Acco dingly, he beha io al pe o mance o child en in he Spanish
asks was o e all mo e a iable han hei pe o mance in he Basque asks (in
phonological and lexical abili ies pa icula ly, see Figu e 2.2). I is also possible ha a
la ge sample could ha e e ealed he signi ican beha io al-CTS ela ionships in
Spanish, by coping wi h his g ea e L2 pe o mance a iabili y.
O e all, he p esen indings in o m he heo e ical, empi ical, and
compu a ional accoun s ha sugges ha p o icien language models play an impo an
ole when ex ac ing linguis ic in o ma ion om con inuous speech. Namely, “adul ”
(o “dominan and p o icien ”) language models can help exploi ing syn ac ic s uc u es
o o ganize linguis ic meaning (Ma in, 2020; Ma in & Doumas, 2017), g ouping wo ds
wi hin o be ween ph ases (Meye e al., 2017), and p edic ing upcoming linguis ic
in o ma ion (Molina o e al., 2021; en Oe e & Ma in, 2021). Specially ele an o ou
indings is he s udy by Kau eld e al. (2020), which aimed o disen angle he ole o
p osody om he ole o linguis ic s uc u e and meaning o del a-band CTS in adul s.
They ound ha linguis ic con en modula ed del a neu al oscilla o y ac i i y beyond
s imulus-d i en p osodic iming. I is he e o e possible ha , du ing language
de elopmen , del a CTS helps map he acous ically d i en p osodic iming while
indi iduals lea n a language and also accumula e linguis ic knowledge p esen in simila
acous ic imescales such as syn ac ic in o ma ion. In e es ingly, Menn, Wa d, e al.
(2022) epo ed ha 10 mon hs-old in an s could ack s essed syllables wi hin del a
equency band and ha hese skills p edic ed ocabula y knowledge a 24 mon hs.
Fu u e longi udinal s udies assessing he di e en eliance on empo al o linguis ic
79
ea u es o acking he speech signal as a unc ion o age and language knowledge, will
help cla i y whe he and when his po en ial de elopmen al shi akes place.
In summa y, ou indings shed new ligh on he ole o AoE and language
knowledge on neu ocogni i e mechanisms suppo ing speech comp ehension such as
CTS. Fo he i s ime, we show ha he co ical acking o he empo al speech
en elope and lexico-syn ac ic speech ea u es does no depend on le els o exposu e, a
leas in ou g oup o bilingual child en. We also highligh ha indi idual di e ences in
speci ic language abili ies depend on he co ical acking o ela ed linguis ic ea u es
con eyed by con inuous speech. In gene al, ou indings can in o m u u e esea ch
abou he ole o linguis ic inpu on CTS, as well as a de elopmen al accoun o which
linguis ic ea u es a e acked by a de eloping neu ocogni e language model.
S udy 2 ocused on in es iga ing he in luence o AoE on CTS. As s a ed in he
Gene al In oduc ion, he e is a ield o esea ch ha ocuses on whe he in an - and
child-di ec ed speech egis e s could be po en ially bene icial o engaging e icien CTS
mechanisms in a de eloping language sys em. The e o e, S udy 3 explo ed he empo al
egula i ies o child-di ec ed speech in o de o de e mine whe he he “quali y” o
speech could be exploi ed by child en while de eloping hei language-speci ic
phonology.
80
3. S udy 3. Local empo al egula i ies in child-
di ec ed speech
2
3.1. In oduc ion
Unde ypical lis ening condi ions, humans e o lessly p ocess and comp ehend
speech as i un olds o e ime. Se e al heo ies sugges ha co ical oscilla ions ( he
ela i ely egula synch onous i ing o neu onal popula ions) in he audi o y and
b oade language egions synch onize o he speech signal a se e al imescales (Ghi za,
2011; Gi aud & Poeppel, 2012). Such synch oniza ion mechanisms allow he empo al
p ocessing o speech and acili a e i s comp ehension. Neu ophysiological esea ch
co obo a es his iew by showing co ical acking o speech acous ic cues ha map
on o linguis ic syllables and p osodic pa e ns (e.g., Ding & Simon, 2012; Doelling e al.,
2014; Molina o & Liza azu, 2018; Peelle e al., 2013). Mo eo e , he e is di ec e idence
ha links an e icien co ical acking o p osodic (Rimmele e al., 2021; del a band
oscilla ions, 0.5 – 4 Hz) and syllabic (Doelling e al., 2014; he a band oscilla ions, 4 – 8
Hz) acous ic cues in he speech signal wi h speech comp ehension. While mos o he
e idence abou he oscilla o y mechanisms o acking acous ic egula i ies in speech
comes om p o icien adul popula ions, in an s’ and child en’s abili ies o ack he
empo al cues o speech ha e also been s udied (e.g. A ahe i e al., 2022; Ge ain &
We ke , 2013; Ríos-López e al., 2017; Tallal, 1980). Howe e , he e is cu en ly li le
2
A simila e sion o his s udy is published as: Pé ez-Na a o, J., Lallie , M., Cla k, C., Flanagan,
S., & Goswami, U. (2022). Local Tempo al Regula i ies in Child-Di ec ed Speech in Spanish. Jou nal o
Speech, Language, and Hea ing Resea ch, 65(10), 3776–3788. h ps://doi.o g/10.1044/2022_JSLHR-22-
00111
81
e idence conce ning whe he he empo al egula i ies o child-di ec ed speech (CDS)
a e enhanced (as compa ed o adul -di ec ed speech, ADS) in o de o suppo and
guide he eme gence o a phonological sys em. The e is also li le e idence conce ning
which s a is ical o ms his empo al enhancemen may ake. Answe ing his ques ion
is c ucial o a comp ehensi e de elopmen al amewo k ha conside s how he b ains
o in an s and child en exploi he empo al egula i ies o he speech hey a e ypically
add essed wi h o achie e p o icien language comp ehension.
Se e al s udies ha e highligh ed he p esence o empo al egula i ies wi hin he
p osodic and syllabic imescales, which in o m he aims o he p esen s udy. A he
syllabic le el, he a e o app oxima ely 5 syllables pe second (5 Hz) is common ac oss
languages (Ding e al., 2017; G eenbe g e al., 2003). A he p osodic s ess le el, Tilsen
and A ani i (2013) showed ha ampli ude en elope-based me hods (simila o hose
used in he p esen s udy) could cap u e s ess egula i ies in spon aneous u e ances.
In he same ein, Inba e al., (2020) ound ha p osodic uni s ( e med ‘in ona ion
uni s’ in hei s udy) p oduced by adul speake s appea a a oughly cons an a e o ~1
Hz. In e es ingly, S ehwien and Meye (2021) analyzed an anno a ed co pus o adio
newscas s in Ge man o show ha he p osody o in ona ional ph ases (mapping on o
u e ances) de e mined he pe iodici y o hei nes ed subo dina e ph ases, sugges ing
ha p osody could ha e a de e mining ole in shaping he local empo al egula i ies o
adul -di ec ed speech. O e all, he e idence sugges s ha he e is a close o e lap
be ween he hy hms o quasi- egula speech uni s such as s essed syllables and
syllables and he imescales a which neu ophysiological mechanisms ope a e o
subse e hei p ocessing (see Poeppel & Assaneo, 2020 o a comp ehensi e e iew on
he hy hms o speech p oduc ion and pe cep ion).
82
While i is well es ablished ha human neu ocogni i e abili ies sub ending he
ex ac ion and segmen a ion o phonological uni s in speech ine- une and gain
language speci ici y du ing he ea ly yea s o li e ( o e iews see Kuhl, 2004; Skeide &
F iede ici, 2016; We ke & Hensch, 2015), i is s ill unclea how he speech inpu s
di ec ed o in an s and child en p o ide hem wi h obus empo al s a is ics ha can
suppo his phonological uning. O pa icula in e es o he p esen s udy a e he
low- equency empo al s a is ics p esen in he ampli ude en elope o he speech
signal, go e ned by ampli ude modula ions (AMs) cen e ed a di e en empo al a es.
AMs a e sys ema ic in ensi y changes in he speech signal, mainly aking place a he
del a (~2 Hz) and he a (~5 Hz) a e bands o AM, ha help o signal he occu ence o
linguis ic uni s like p osodic ph asing (~1000 ms) and syllables (~200 ms) espec i ely
(Ding, Pa el, e al., 2017; G eenbe g, 2006; G eenbe g e al., 2003). Such empo al
luc ua ions in he ampli ude en elope o he speech signal, pa icula ly he AM ise
imes ( a es o change o hese AM bands), p o ide salien acous ic ma ke s ele an
o ex ac ing p osodic and syllabic phonological uni s, while as e modula ions (~35
Hz) a e hough o con ibu e o he ex ac ion o phonemic in o ma ion (Poeppel e
al., 2008). The iden i ica ion o phonological uni s in he speech signal is c ucial o
phonological and eading de elopmen (Ziegle & Goswami, 2005). Beha io al
e idence, in line wi h he e idence on he co ical acking o speech (e.g., Doelling e
al., 2014; Rimmele e al., 2021), highligh s he unc ional ole o acking del a and he a
AMs in sen ence segmen a ion and syllabic pa sing espec i ely (Ghi za, 2012, 2017). A
key unc ional ole o del a and he a AMs is also in line wi h he empo al sampling
heo y (Goswami, 2011), a de elopmen al amewo k o language acquisi ion cen e ed
on phonology. The empo al sampling heo y p oposes ha he au oma ic alignmen o

83
endogenous b ain hy hms wi h AM-go e ned hy hm pa e ns in speech is c i ical o
linguis ic and phonological de elopmen , and ha his unconscious neu al alignmen
(o sampling) p ocess may be a ypical in de elopmen al dyslexia, which is cha ac e ized
by bo h phonological and ampli ude ise ime di icul ies.
Cohe en wi h he empo al sampling hypo hesis, wo bodies o e idence a es
he key ole o acking low equency speech AMs o phonological de elopmen .
Fi s ly, mul iple s udies ac oss languages ha e shown ha impai men s in AM
sensi i i y accompany he a ypical phonological de elopmen cha ac e is ic o
de elopmen al dyslexia (e.g., Goswami e al., 2002, 2010; Leong & Goswami, 2014;
Su ányi e al., 2009; see Hämäläinen e al., 2012 o a sys ema ic e iew). Secondly,
sensi i i y o AMs du ing he i s yea s o li e is a p edic o o ou comes in undamen al
language domains, such as phonological awa eness (Goswami, Wang, e al., 2010;
Van oo en e al., 2017), ocabula y (Kalashniko a e al., 2019), and eading abili ies
(Van oo en e al., 2017). In addi ion, ecen longi udinal s udies show ha co ical
oscilla o y acking o p osodic in o ma ion is p esen in in an s om 4 mon hs, and
inc eases du ing ea ly childhood (Ríos-López e al., 2020; A ahe i e al., 2022),
sugges i e o he ele ance o del a-band speech acking o language de elopmen .
Ríos-López e al., (2021) showed ha a bigge del a-band co ical acking o speech in
p e- eading child en indeed p edic s be e eading skills one yea la e , a e he
beginning o o mal eading ins uc ion.
P e ious e idence shows ha adul s adap hei speech complexi y o child en’s
linguis ic abili ies and communica i e eedback, in o de o acili a e comp ehension
(Hu enloche e al., 2010; Kalashniko a e al., 2020; Lam & Ki amu a, 2012; Smi h &
84
T aino , 2008). The e is abundan e idence conce ning he spec al (pi ch)
cha ac e is ics o in an -di ec ed speech (IDS), which a e exagge a ed o make i a
phone ically-salien and engaging egis e o add ess language-lea ning indi iduals
(Dilley e al., 2020; Fe nald, 1985; Kuhl e al., 1997; T aino & Desja dins, 2002; We ke
e al., 2007; We ke & McLeod, 1989; see Fe nald, 2000 o a e iew). The enhanced
spec al cha ac e is ics o IDS a e well-es ablished, howe e less is known ega ding
po en ial empo al adap a ions ha may ake place when add essing in an s and
child en. Two well-known empo al ea u es o IDS a e a slowe speech a e and sho e
u e ances (Fe nald e al., 1989; Fe nald & Simon, 1984; Leong e al., 2017). I may be
he case ha CDS could also p o ide especially egula empo al s a is ics o acili a e
iden i ica ion o and access o phonological uni s in speech and he eby o acili a ing
he eme gence o a p o icien phonological sys em. Such a hypo hesis was ini ially
explo ed by Leong and Goswami (2015) in ela ion o he AM o ganiza ion o CDS in
English, ypically ega ded as a s ess- imed language (i.e., a language cha ac e ized by
ce ain egula i y in he iming o s essed syllables); and u he es ed by con as ing
IDS and ADS in English (Leong e al., 2017). In he la e s udy, Leong e al. (2017)
showed ha IDS di e ed om ADS in i s empo al o ganiza ion, especially ega ding
wo c i ical aspec s. One was he highe p ominence o del a band modula ion ene gy
in IDS compa ed o ADS: he modula ion spec um e ealed ela i ely mo e powe in
he del a band o IDS han o ADS. This ea u e is likely linked o enhanced p osody
in IDS, p o iding mo e salien empo al in o ma ion ele an o ex ac ing phonological
in o ma ion a slowe imescales (e.g., in ona ion ph ases, wo ds, and s essed syllables)
o a lea ning indi idual. The second ea u e was ha s essed syllables we e mo e
egula ly spaced in IDS han ADS, shown by signi ican ly g ea e phase synch oniza ion
85
( hy hmic alignmen ) o del a- a e and he a- a e AMs (~2Hz and ~5 Hz espec i ely)
in IDS. This was in e p e ed as p o iding a p edic able empo al skele on o acili a e
he in an ’s a en ional and pe cep ual access o syllables du ing ea ly s ages o language
lea ning.
Howe e , o da e, he e is no s udy conce ning he po en ial bene i ha he
empo al o ganiza ion o CDS (in con as o ADS) could p o ide du ing p e-school
yea s, no o wha ex en such empo al o ganiza ion is p esen in non-s ess imed
languages like Spanish. Languages like Spanish a e cha ac e ized by salien syllabic
iming, and hus ha e been adi ionally ca ego ized as syllable- imed languages, (see
Ramus e al., 1999, and Va ne e al., 2017, o ins ances o suppo ing e idence; bu also
A ani i, 2009; Tu k & Sha uck-Hu nagel, 2013 o opposing iews). He e we ocus on
kinde ga en, a s age in which phonological abili ies (e.g., phonological awa eness and
phonological sho - e m memo y) a e explici ly augh , as hey will suppo la e
eading acquisi ion (e.g., Ca a olas e al., 2001; Mu e e al., 2004). We in es iga ed
whe he he empo al egula i ies o CDS di e ed om hose o ADS in Spanish, by
di ec ly con as ing he empo al s a is ics o he wo speech egis e s wi hin he same
s udy o he i s ime. I CDS shows simila salien empo al ea u es o English, in
p inciple his could signal he p esence o language-uni e sal empo al s a is ics ha
may acili a e lea ning, pa icula ly ega ding an eme gen phonological sys em. To his
pu pose, we ocused on h ee empo al ea u es o speech: he modula ion spec um,
he empo al egula i y o he placemen o s essed syllables and syllable a e. We
s udied he wo ea u es —modula ion spec um and he empo al egula i y o he
placemen o s essed syllables— ha Leong e al., (2017) al eady ound dis inc i e in
IDS in English, a s ess- imed language. The modula ion spec um o each speech
86
egis e was compu ed and he a ea unde he cu e (AUC) was compa ed in del a
e sus he a bands o CDS and ADS espec i ely. Ou aim was o disce n whe he in
Spanish, he wo speech egis e s can be di e en ly ca ego ized as mo e p osody-salien
(g ea e AUC in he del a- a e AM band) o syllable-salien (g ea e AUC in he he a-
a e AM band). To cha ac e ize he egula i y wi h which syllables we e s essed in CDS
in con as o ADS, we analyzed he empo al alignmen be ween del a and he a AM
bands in e ms o AM phase alignmen ( hy hmic synch onici y). To his pu pose, we
used he spec al-ampli ude modula ion phase hie a chy (S-AMPH) model de eloped
by Leong and Goswami (2015). The S-AMPH model allows us o decompose he
ampli ude en elope o he speech signal and measu e he empo al alignmen be ween
di e en AM bands nes ed wi hin he signal in di e en wo ds and ph ases in e ms o
hei phase synch oniza ion (see Figu e 3.1 o a ph asal example). O pa icula in e es
o ou s udy, del a- he a phase alignmen plays a c ucial ole in he pe cep ion o
p osodic pa e ns in English and has been p oposed as a no el s a is ic o he language-
lea ning b ain (Leong & Goswami, 2015). G ea e del a- a e o he a- a e AM phase
synch oniza ion is hough o help o iden i y p osodic pa e ning by speci ying s ong
e sus weak syllables (Leong e al., 2014). When bo h AM bands peak oge he , a s ong
syllable is hea d. When a ough in he slowe del a- a e AM band (cen e ed on ~2 Hz in he
speech ma e ials used by Leong e al., 2014) coincides wi h a peak in he as e he a- a e AM
band (cen e ed on ~4 Hz in Leong e al., 2014), a weak syllable is hea d. Whe he he same is
ue in Spanish is cu en ly unknown. Finally, we analyzed syllable a e. Ou goal was o
ex end p e ious indings o CDS being mo e slowly paced han ADS (Bie sack e al.,
2005; Sjons e al., 2017), and o in es iga e he po en ial links be ween a pu a i e slowe
speech a e in CDS and i s expec ed enhanced empo al egula i ies. In summa y, he
93
au oma ic syllable a e me ics wi h hei co esponding manually anno a ed syllable
a e indexes, es ima ed by ained na i e speake s, showing indeed a high co ela ion
be ween manually anno a ed and au oma ically de ec ed syllable a e ( (1582) = .95, p
< .001; Supplemen al igu e 3.2).
3.3. Resul s
In o de o assess he in luence o speech egis e (CDS, ADS), and speaking
condi ion (spon aneous speech, ead speech) on each o ou empo al measu es
(dis ibu ion o modula ion ene gy, phase synch oniza ion, and syllable a e), we used
linea mixed e ec (LME) models. Gi en he wi hin-pa icipan s uc u e o ou s udy,
we included each pa icipan as a andom in e cep in he model. We used he lme
unc ion o lme4 package ( .1.1.28, Ba es e al., 2015) as well as ano a unc ion o es
he omnibus main e ec s and in e ac ions o ou p edic o s.
3.3.1. Modula ion Spec um (P osodic salience)
To ope a ionalize ou planned analyses conce ning he peak loca ions o he
modula ion spec a (Figu e 3.2, panel A), we calcula ed he a ea unde he cu e (AUC),
de ined as he linea ans o ma ion o each equency band’s di e ence in dBs om
mean powe . Del a and he a segmen s o he modula ion spec um di e ed g ea ly in
hei AUC (Figu e 3.2, panel A), as p e iously shown by o he s udies (e.g., Ding, Pa el,
e al., 2017). O e all (i.e., ac oss egis e s and condi ions), AUC was signi ican ly bigge
in he a han in del a, (36) = 25.62, p < 0.001 (β = 0.260, SE = 0.010, CI [0.240 0.280]).
The e o e, we ci cumsc ibed ou planned analyses o each o he AM bands sepa a ely.
LME showed ha , wi hin del a, he e we e signi ican e ec s o speaking egis e , F(1,

94
54) = 11.45, p = 0.001, condi ion, F(1, 54) = 51.43, p < .001, and an in e ac ion be ween
egis e and condi ion, F(1, 54) = 23.39, p < 0.001. This pa e n o esul s e eals a bigge
del a AUC in CDS han in ADS, (54) = 5.81, p < 0.001 (β = 0.066, SE = 0.011, CI [0.043
0.089]), as well as in spon aneous han in ead speech, (54) = 8.49, p < 0.001 (β = 0.097,
SE = 0.011, CI [0.074 0.120]). In he he a segmen o he modula ion spec um, LME
also yielded a signi ican e ec o speaking egis e , F(1, 54) = 20.51, p < 0.001, condi ion,
F(1, 54) = 98.07, p < 0.001, as well as an in e ac ion be ween bo h ac o s, F(1, 54) =
18.79, p < 0.001. Howe e , he he a segmen o he modula ion spec um was
cha ac e ized by he in e se pa e n ela i e o del a, namely ADS showing a bigge
he a AUC han CDS, (54) = 6.27, p < 0.001 (β = 0.027, SE = 0.004, CI [0.019 0.036]),
as well as ead speech showing a bigge he a AUC han spon aneous speech, (54) =
10.07, p < 0.001 (β = 0.044, SE = 0.004, CI [0.035 0.052]). Indeed, o he a, he
modula ion spec um o all condi ions peaked a a ound 5 – 6 Hz, co esponding o he
syllable a e (as p e iously shown ac oss languages; Ding, Pa el, e al., 2017; G eenbe g
e al., 2003).
95
Figu e 3.2. A) Modula ion spec a o he ou speaking condi ions. The e ical g ey lines di ide
he signal-de i ed modula ion a es o he S-AMPH model ha we used o de ine del a and
he a bands, o which we subse ou PSI and AM spec um analyses. B) A ea unde he cu e
(AUC) o he del a (le ) and he a ( igh ) bands o spon aneous CDS, ead CDS, spon aneous
ADS, and ead ADS espec i ely om le o igh . The ho izon al lines be ween condi ions
ep esen signi ican di e ences in AUC, adjus ed o mul iple compa isons (** p < .01; **** p
< .0001). C) AUC o canonical he a band (4 - 7 Hz). Signi ican di e ences be ween speaking
condi ions a e ep esen ed as in sec ion B. Do s in Panels B and C ep esen mean AUC alues
pe pa icipan .
Thus, spon aneous and ead CDS had signi ican ly g ea e modula ion ene gy
(i.e., bigge del a AUC) han spon aneous and ead ADS espec i ely, sugges i e o mo e
96
salien p osodic s uc u e in CDS. The esul s o spon aneous speech a e in line wi h
he IDS-ADS p osodic di e ences in IDS in English demons a ed by Leong e al. (2017).
The da a o ead CDS a e comple ely no el. Mo eo e , and in line wi h he di e ences
be ween ead and spon aneous ma e ials ha ha e been epo ed wi h espec o
p osody (e.g., Hi ose & Kawanami, 2002; Howell & Kadi-Hani i, 1991), ou esul s
sugges ha when eading o o spon aneously add essing adul s in a syllable- imed
language, a g ea e syllabic salience akes place (i.e., bigge he a AUC).
3.3.2. Regula i y o s essed syllables (del a- he a phase synch oniza ion,
PSI)
Del a- he a PSI alues in he di e en spec al bands demons a ed a simila
pa e n ac oss speech egis e s (Supplemen al Figu e 3.3). The e o e, we i s compu ed
an LME model wi h mean PSI alues as he dependen a iable. The LME yielded a
signi ican e ec o speaking egis e , F(1, 54) = 26.82, p < .001, showing ha CDS is
cha ac e ized by highe del a- he a phase synch oniza ion han ADS, (54) = 2.49, p =
0.016 (β = 0.011, SE = 0.004, CI [0.002 0.020]) (Figu e 3.3). The e was no signi ican
e ec o speaking condi ion (spon aneous s. ead) no in e ac ion be ween egis e and
condi ion (p > .05).
97
Figu e 3.3. Mean del a- he a PSI. G ay lines connec pa icipan s’ mean PSI ac oss condi ions.
Ho izon al lines wi hin each box ep esen median PSI. Uppe and lowe hinges ma k he i s
and hi d qua ile, and whiske s show 1.5 * in e -qua ile ange. Bon e oni-co ec ed signi ican
di e ences a e ep esen ed wi h * (p < .05) and ** (p < .01).
3.3.3. Syllable a e
The LME model on syllable a e yielded signi ican e ec s o speech egis e , F(1,
54) = 8.32, p = .006, and condi ion, F(1, 54) = 7.93, p = .007, bu no in e ac ion be ween
hese ac o s (p > 0.05). These main e ec s a e isible in Figu e 3.4, which shows he
highe syllable a e o ADS ela i e o CDS, (54) = 2.203, p = 0.032 (β = 0.262, SE =
0.119, CI [0.025 0.499]), and o ead speech ela i e o spon aneous speech, (54) =
2.155, p = 0.036 (β = 0.256, SE = 0.119, CI [0.019 0.493]). Figu e 3.4 also shows ha he e
is much less a iabili y in he speech a e o ead speech, and in e es ingly, pa icula ly
o speech ead o child en (CDS). This sugges s ha eade s spon aneously adap hei
98
speech when eading o child en o make i highly p edic able. I should be no ed ha
he me hod we used o calcula e syllable a e yields sligh ly smalle alues han manual
anno a ion o o he ypically used calcula ions. Acco dingly, we mul iplied ou syllable
a e alues by 1.28 as s a ed in he me hod’s manusc ip (de Jong & Wempe, 2009). This
con i med an o e lap wi h he peak o he modula ion spec um in he he a band o
each egis e and condi ion (Figu e 3.2).
Nex , we analyzed he ela ionship be ween syllable a e and he empo al
egula i y o he u e ances. The nega i e co ela ions be ween syllable a e and del a-
he a PSI we e signi ican (Figu e 3.5). This shows ha he slowe paced u e ances we e
he mos empo ally o ganized u e ances in ou da ase .

99
Figu e 3.4. Syllable a e ac oss speaking condi ions. G ay lines connec pa icipan s’ mean
syllable a es ac oss condi ions. Ho izon al lines wi hin each box ep esen median syllable a e.
Uppe and lowe hinges ma k he i s and hi d qua ile, and whiske s show 1.5 * in e -qua ile
ange. Bon e oni-co ec ed signi ican di e ences a e ep esen ed wi h * (p < .05).
Figu e 3.5. Co ela ion be ween syllable a e and del a- he a PSI. The ou lines indica e he
slopes o i ed linea models o each speaking condi ion. Top le : Pea son co ela ion
coe icien s and p- alues o each speaking condi ion.
3.4. Discussion
In he p esen s udy, we in es iga ed bo h spon aneous and ead CDS and ADS
in Spanish wi h he objec i e o con as ing hem in e ms o empo al egula i ies. Ou
100
wi hin-pa icipan design allowed us o in es iga e whe he adul s lexibly adap hei
spon aneous speech p oduc ions o boos speech empo al egula i ies when add essing
4-yea -old child en a he han o he adul s. Using h ee empo al me ics, we ound
ha CDS in Spanish ca ies mo e egula empo al s a is ics han ADS. Fi s , CDS has
signi ican ly mo e modula ion ene gy in he del a band han ADS, whe he i is spoken
spon aneously o whe he he adul is eading o he child. Second, CDS con ains mo e
egula ly s essed syllables han ADS, as shown by he g ea e phase alignmen o he
del a- a e and he a- a e AM bands in he CDS egis e s. Thi d, CDS shows a slowe
syllable a e ela i e o ADS. Adul s slow down when speaking o child en, as migh be
expec ed when add essing language-lea ning indi iduals. Addi ionally, ead CDS also
showed a no ably na owe ange han he o he egis e s ega ding syllable a e,
sugges ing ha when eading o young child en, empo al in o ma ion becomes highly
p edic able. This may help o explain why ea ly s o y eading is such an impo an
con ibu o o language de elopmen (A ig & Weine , 2020).
The ampli ude modula ion spec um o CDS sugges ed ha i has signi ican ly
mo e modula ion ene gy in he del a band han ADS (Figu e 3.2). This is in line wi h
p io IDS da a in English (Leong e al., 2017), classically conside ed a s ess- imed
language. The ac ha we also ound enhanced p osodic salience in CDS in Spanish,
ypically e med a syllable- imed language, is consis en wi h he idea ha IDS and CDS
boos ela i ely slow sup asegmen al in o ma ion o aid he mapping o phonological
uni s by language-lea ning indi iduals (Fe nald, 2000). Neu ophysiological s udies
(including S udy 2) show ha in an s and child en ely on sup asegmen al/p osodic
han syllabic in o ma ion o acking and segmen ing con inuous speech (A ahe i e
al., 2022; Ríos-López e al., 2020), which may help o explain he enhanced del a band
101
modula ion ene gy in Spanish CDS. Despi e his enhancemen o del a-band
modula ions, and as expec ed, he modula ion spec um o his syllable- imed
language peaked in he he a band o bo h Spanish CDS and ADS, as has p e iously
been epo ed ac oss languages o ADS (e.g., Ding, Pa el, e al., 2017; G eenbe g e al.,
2003). Howe e , ADS showed signi ican ly mo e modula ion ene gy in he he a band
compa ed o CDS. This migh indica e ha he empo al egula i ies o ADS a e mo e
sys ema ic a he syllabic imescale, cohe en wi h he e idence o syllables being a
undamen al empo al landma k o adul neu ocogni i e speech p ocessing abili ies
(Doelling e al., 2014; Ghi za, 2012).
Ou indings o highe del a- he a PSIs in CDS sugges ha i s s essed syllables
a e empo ally mo e egula ly placed han in ADS. Indeed, p io speech modelling wo k
has shown ha del a- he a AM phase ela ionships unde pin speech hy hm pe cep ion
(Leong e al., 2014), wi h AM peak synch oniza ion helping o de e mine he pe cei ed
me ical pa e ning o u e ances such as ochaic e sus iambic. Gi en he syllable-
imed na u e o Spanish, he g ea e p edic abili y o s essed syllables may help he
phonological mapping o Spanish by language-lea ning indi iduals. Ou da a hus
con ibu e o he cu en e idence on con inuous speech hy hmici y, by showing ha ,
a u e ance le el, CDS is mo e hy hmic han ADS. These indings a e also in line wi h
p e ious adul s udies ha ha e con ex ualized he empo al egula i ies o speech
wi hin local (u e ance le el) s ess pa e ns (A ani i, 2009; Nolan & Jeon, 2014; Tilsen
& A ani i, 2013). Indeed, he e is ecen e idence o local p osodic s ess egula i ies
in ADS in di e en languages (e.g., Inba e al., 2020; S ehwien & Meye , 2021). The
slowe syllable a e in CDS ela i e o ADS is also o ele ance when compa ing empo al
s a is ics. In summa y, CDS appea s o o e a con inuous speech s eam ha is easie
102
o segmen ia slowe speech a e ( ewe syllables pe second), g ea e hy hmici y
(p edic abili y o occu ence o s essed syllables), and he enhancemen o del a-band
speech in o ma ion (p osody-salien egis e ). In line wi h his in e p e a ion, p e ious
e idence shows ha while adul neu ocogni i e mechanisms adap o di e en speech
a es wi hin he 4 – 7 Hz syllabic ange (e.g., Foulke & S ich , 1969; Ghi za, 2011; Liza azu
e al., 2019), child en’s comp ehension abili ies bene i om slowe speech a es (Be y
& E ickson, 1973; Haake e al., 2014; Mon gome y, 2004; Riding & Vincen , 1980). The
adap ed empo al s a is ics in Spanish CDS demons a ed he e could hus aid
comp ehension by child en as well as acili a e he de elopmen o a phonological
sys em. The enhanced local (u e ance-le el) empo al egula i ies o CDS, whe he i is
ead o spoken, p o ide a se o empo al s a is ics ha can be exploi ed by child en’s
neu ocogni i e mechanisms o s a is ical (Rombe g & Sa an, 2010) and dis ibu ional
lea ning (see Banai & Ahissa , 2018 o a e iew). Sensi i i y o hese AM- ela ed
s a is ics would enable a child o build inc easingly mo e obus phonological
ep esen a ions a wo d and syllable le el. Indeed, he mapping o speech empo al
s a is ics is known o be ine icien in indi iduals wi h phonological de ici s such as
dyslexia (Ahissa e al., 2006; Banai & Ahissa , 2018; Goswami, 2011; Leong & Goswami,
2017). P e ious s udies wi h adul s ha e also shown p osodic (Inba e al., 2020) and
syllabic (Ding, Pa el, e al., 2017) egula i ies ac oss languages. Howe e , ou esul s
highligh ha g ea e empo al synch oniza ion be ween del a- a e and he a- a e AMs
may be a speci ic cha ac e is ic o CDS (in con as o ADS). This inding is consis en
wi h an ‘acous ic-eme gen ’ pe spec i e ega ding phonological de elopmen om
in ancy onwa ds (Leong & Goswami, 2015).
109
been epea edly linked o lexical and syn ac ic de elopmen abo e and beyond he me e
epea ed exposu e o a language (Fu ow e al., 1979; Ho -Ginsbe g, 1986; Hu enloche
e al., 2002, 2010).
Second, ano he se o indings suppo s he ole o AoE on language
de elopmen showing ha phonological and non-phonological abili ies di e en ely
on exposu e in o de o de elop. O e all, we could no highligh any di ec in luence o
AoE on phonological abili ies (phonological sho - e m memo y) in any o he
languages o he bilinguals, al hough we did epo a language e ec on pe o mance:
h oughou he longi udinal assessmen , child en epea ed nonwo ds ha con ained
lexical s ess and syllabic phono ac ic cha ac e is ics o Basque, hei dominan
language, compa ed o Spanish. These indings poin a an indi ec e ec o AoE (i.e.,
media ed by language dominance) on he de elopmen o wo d-le el phono ac ics
(Messe e al., 2010; Munson, 2001; Vi e i ch & Luce, 2005). Such indi ec e ec migh
possibly be media ed by enhanced lexical knowledge in he dominan language, which
we showed was i sel di ec ly in luenced by AoE. In e es ingly, he language e ec on
phonological abili ies was no ound in S udy 2 (unlike p oduc i e ocabula y and
eading p o iciency), when he child en we e a ound 7 yea s o age, and had a highly
unbalanced p o ile in e ms o AoE o hei languages. The lack o language di e ences
on phonological pe o mance migh e lec he gap be ween languages closing a his
de elopmen al s age (see Phonology in Supplemen a y Figu e 2.3). Impo an ly, a his
s age, child en we e also s a ing o acqui e eading skills, which we know exe s an
in luence on phonological de elopmen (Cas les and Col hea , 2004). Thus, acqui ing
eading may ha e igge ed a boos in he de elopmen o he phonological skills o he
child en o S udy 2. In suppo o his hypo hesis, CTS a he en elope le el, ha is

110
hough o s ongly ela e o phonological abili ies (see also Di Libe o e al., 2018), was
no signi ican ly di e en be ween Basque and Spanish. The e o e, he e is he
possibili y ha , a ound 6-7 yea s o age, phonological abili ies in bilinguals a e simila ly
de eloped in bo h languages, wi h he less dominan language eaching na i e-like
le els despi e conside ably limi ed inpu . These indings in o m on how an L2 migh
de elop, gi en he ele ance o phonological abili ies o ex ac ing linguis ic
in o ma ion om he speech signal a ela i ely ea lie s ages o language lea ning (S. E.
Ga he cole, 2006). Gi en he a es ed boo s apping o phonological abili ies o de elop
b oade language abili ies du ing childhood (An hony e al., 2003; Ca oll e al., 2003;
Kehoe e al., 2020; Vaah o an a e al., 2020), na i e-like de eloped phonological skills
could suppo he g ow h o L2 abili ies ha a e no ye a he same s age as in L1 (e.g.,
lexical and syn ac ic abili ies in S udy 1, and lexical and eading abili ies in S udy 2).
Unlike phonological skills, lexical ( ecep i e ocabula y) and syn ac ic (sen ence
epe i ion) ecep i e abili ies g ea ly elied on quan i a i e cha ac e is ics o he speech
inpu . This was especially no iceable in he less dominan language o bilinguals
(Spanish) o which child en imp o ed hese abili ies a a as e a e han in Basque. By
he end o S udy 1 (6 yea s old), child en’s pe o mance in ecep i e ocabula y was
simila in bo h languages, and o sen ence epe i ion, pe o mance was e en highe in
Spanish han Basque. As men ioned abo e, he de elopmen o ecep i e abili ies may
be highly sensi i e o any p esen ed linguis ic inpu s o help boo s apping he
de elopmen o o he language abili ies when AoE is e y limi ed. Findings om S udy
2 suppo his iew. We included child en who we e e y unbalanced bilinguals in e ms
o AoE (mo e dominan in Basque han Spanish) and p oduc i e ocabula y (highe in
Basque han in Spanish). Ye , he neu ocogni i e abili ies ha we know suppo speech
111
comp ehension, and indexed by CTS me ics, we e highly simila ac oss he wo
languages. Like language ecep i e abili ies, CTS in bilingual lea ne s migh suppo
language acquisi ion when he inpu is limi ed by ac ing as an “inpu -hung y”
con inuous speech acking de ice ope a ing independen ly o language dominance and
p o iciency.
Indeed, ou hypo hesis ha la ge language exposu e and knowledge would
esul in mo e e icien neu al acking a he en elope and linguis ic le els was no
suppo ed by ou indings, a leas when measu ed a his s age o de elopmen . This
could be he esul o an in e play be ween exposu e and b ain ma u a ional ac o s in
bilinguals. Linguis ic exposu e in he con ex o bilingual language acquisi ion seems o
le e age on neu oplas ici y o ex end he c i ical pe iods in which languages a e lea n
a na i e-like le els (We ke & Hensch, 2015; Xue e al., 2021). In he case o S udy 2, as
child en had s a ed acqui ing bo h o hei languages om ea ly in li e, he e is he
possibili y ha hey we e s ill wi hin a sensi i e pe iod o de eloping hei
neu ocogni i e abili ies o language comp ehension in bo h o hei languages wi h
na i e p o iciency (as i is he ypical bilingual lea ning p o ile in he Basque Coun y).
Cohe en wi h his in e p e a ion, a he han AoE, he age o acquisi ion o a language
(< 3 y.o. o bo h languages) migh ac as a pe asi e de e minan o he s uc u al and
unc ional e ec s o bilingualism on he neu oplas ici y o language- ela ed b ain a eas
(Be ken, Chai, e al., 2016; Be ken, G acco, e al., 2016; Claussenius-Kalman e al., 2020;
Gulli e e al., 2018; Klein e al., 2014; see Claussenius-Kalman e al., 2020 o an
o e iew o signi ican and null e ec s o age o acquisi ion on b ain s uc u e).
112
Al hough AoE did no modula e he e iciency o he co ical acking o he
dis ibu ional p ope ies o acous ic and linguis ic speech ea u es, we showed ha
indi idual a ia ions in language pe o mance we e speci ically ied o he neu al
acking o he co esponding speech ea u es in con inuous speech: in o he wo ds,
phonological abili ies we e posi i ely ela ed o he acking o ampli ude modula ion
changes in he speech signal (en elope acking), and lexical abili ies we e associa ed o
he co ical acking o (lexical and syn ac ic) linguis ic in o ma ion. The ela ionship
be ween phonological abili ies and he acking o he speech en elope is b oadly in line
wi h he empo al sampling hypo hesis (Goswami, 2011, 2017), ha p oposes a causal
pa hway be ween an e icien co ical acking o he ampli ude modula ions o speech
and phonological (as well as eading) de elopmen . While p e ious s udies ha e shown
ha CTS is a ele an p edic o o ocabula y du ing in ancy (K. Menn, Wa d, e al.,
2022), we now show ha CTS and phonological abili ies a e posi i ely ela ed a age 7,
co esponding o he ea ly s ages o eading acquisi ion. This is cohe en wi h a ecen
s udy showing ha CTS in p e- eade s p edic ed hei eading abili ies a 7 yea s o age,
a he end o hei i s yea o o mal eading acquisi ion (Ríos-López e al., 2021). I is
wo h no ing ha such phonology-en elope CTS ela ionship was only p esen in
Basque (L1), esul ha we did no p edic . Fu he esea ch is he e o e needed o help
elucida e he de elopmen al ange a which CTS is c ucial o phonological and eading
abili ies.
In ela ion o he empo al sampling amewo k o phonological de elopmen ,
S udy 3 showed ha CDS was cha ac e ized by enhanced and mo e salien empo al
dis ibu ional p ope ies o speech, as opposed o ADS. In addi ion o con eying a
ela i ely slowe speech a e han ADS, we obse ed ha CDS was cha ac e ized by
113
enhanced empo al egula i ies a bo h he p osodic (modula ions a he del a
equency ange) and syllabic ( he a equency ange) le els. In e es ingly, S udy 2
showed ha he speech-b ain a bo h del a and he a equencies (as well as en elope-
le el mTRF) was posi i ely ela ed o phonological abili ies. Al oge he , hese indings
b ing con e ging e idence o he ole o slow speech empo al imescales o he
acking o p osodic and syllabic in o ma ion and he build-up o e ined phonological
ep esen a ions. I is wo h no ing ha syllabic acking migh play a mo e impo an
ole o phonological acquisi ion la e in de elopmen as ou 7-yea -old pa icipan s
only showed signi ican del a (and no he a) speech-b ain cohe ence a he g oup le el
(see also Ríos-López e al., 2021). Impo an ly, S udy 3 shows ha adul s adap he
empo al dis ibu ional s a is ics o hei speech u e ances when add essing language-
lea ning indi iduals, o ma ch he capaci y o hei de eloping neu ocogni i e
p ocessing sys em o speech. Ideally, u u e de elopmen al esea ch going beyond he
age ange o ou s udy could help delinea ing whe he CDS is less p oduced by adul s
as child en become mo e p o icien in hei language(s).
Impo an ly, ou esul s om S udy 2 show ha child en wi h a easonably high
language expe ise (7 yea s old) migh ex ac meaning om speech h ough he
acking o empo al dis ibu ional p ope ies o speech ha go beyond pu ely acous ic
ea u es. In ac , S udy 2 shows a obus ela ionship be ween he co ical acking o
“linguis ic” (lexical and syn ac ic) dis ibu ional speech p ope ies and ocabula y
knowledge. S udy 2 is o ou knowledge he i s e idence o he occu ence o such
linguis ic neu al acking in a child popula ion. While many s udies in in an s and
child en (including S udy 2) highligh he ele ance o en elope-le el (acous ic) CTS o
de eloping phonological and b oade language abili ies (e.g., ocabula y, eading; e.g.,
114
A ahe i e al., 2022; K. Menn, Wa d, e al., 2022; Ríos-López e al., 2020, 2021), only
adul s udies had highligh ed so a ha CTS goes beyond acking he pu e acous ics
o he signal (B ode ick e al., 2021; Klimo ich-G ay e al., 2021; Meye e al., 2017;
Molina o e al., 2021). The “cos -minimiza ion” a gumen p oposes ha once
indi iduals each enough language knowledge, hei p o icien neu ocogni i e language
models do no need o ac i ely ex ac linguis ic in o ma ion om he speech signal by
elying on phonological abili ies. Ins ead, p o icien language use s can exploi hei
language knowledge o p edic upcoming in o ma ion based on hei lexico-seman ic
and syn ac ic knowledge (Ma in & Doumas, 2017; en Oe e & Ma in, 2021). This
obse a ion could be he i s documen a ion hin ing a a de elopmen al shi a ec ing
CTS: om he acking o pu ely acous ic in o ma ion o ma ch he equi emen s o an
imma u e language sys em, o he acking o highe o de linguis ic in o ma ion when
he sys em has eached su icien language knowledge. None heless, such
de elopmen al shi migh s a much ea lie han a ound 7 y.o. (i.e., when child en can
comp ehend he as majo i y o he wo ds in he speech hey a e exposed o, a ound
3-4 yea s o age), which is an in e es ing ques ion o be add essed by u u e
de elopmen al esea ch.
Limi a ions
Se e al limi a ions o his doc o al wo k ha e o be aised. The i s one is ela ed
o sample size. Ou s udies could be seen as “medium sample size” (N = 74, 35, and 18
o S udy 1, 2, and 3 espec i ely). These sample sizes we e cons ained by
me hodological and easibili y ac o s, as S udy 1 was longi udinal and included speech
co po a analyses, in S udy 2 combined EEG and beha io al assessmen s, and S udy

115
included speech manual ansc ip ions and analyses. Fu u e s udies could ely on mo e
b oadly au oma ized p o ocols o assessing language pe o mance in spon aneous
p oduc ions (e.g., au oma ic ansc ip ions, na u al language p ocessing ools o
language de ec ion and analysis). This will enable esea che s o include la ge samples
o a mo e obus es ima ion and analyses o de elopmen al da a, which is p one o be
in luenced by nume ous sou ces o he e ogenei y. Ano he limi a ion o he
gene alizabili y o ou indings is he ac ha we did no include s anda dized no
composi e sco es o se e al o he language measu es (e.g., phonological abili ies),
which limi s he obus ness wi h which hey a e es ima ed. This issue a ises om
including a wide se o measu es in bo h languages, which limi s he use o a common
s anda dized ba e y as well as o se e al measu es o he di e en domains o c ea e
composi e sco es. Ne e heless, u u e esea ch could ocus on speci ic ela ionships
e ealed by his hesis and es he eliabili y o ou indings wi h mo e obus es ima es.
Ano he limi a ion is he ac ha we did no include a con ol condi ion in S udy 2,
which could consis in es ing he co ical acking o an unknown language o o
ampli ude-modula ed noise in o de o con as he impac o in elligibili y (o lack
he eo ) on CTS. Ano he speci ic limi a ion o S udy 3 is he ac ha we es ed he
spon aneous p oduc ions o pa icipan s when add essing hei child en (CDS) and
unknown adul s (ADS). Such di e ence in amilia i y wi h he add essees migh ha e
con ibu ed o a mo e engaging CDS (in con as o ADS) and possibly boos he CDS-
ADS di e ences.
116
Conclusion
The o e a ching ques ion o he p esen hesis was whe he he quan i y and
quali y o linguis ic inpu play a ele an ole in language de elopmen du ing
childhood. We showed ha he de elopmen o speci ic non-phonological language
domains, such as lexico-seman ics and syn ax, a e di ec ly impac ed by he amoun o
linguis ic exposu e. On he opposi e, phonological abili ies as well as he co ical
acking o speech we e no shown o depend on AoE. Impo an ly, we demons a e
ha , when assessing exposu e ac o s modula ing language g ow h, he s udy o
language dominance and p o iciency is complemen a y o he in es iga ion o mo e
di ec con inuous measu es o pe cen age o AoE.
Wi h espec o he quali y o linguis ic exposu e, we showed ha , when alking
o hei child en, adul s adap hei speech egis e o boos speech empo al
egula i ies, as a po en ial s a egy o mee he neu ocogni i e equi emen s o he
imma u e language p ocessing sys em o young lea ne s.
In closing, we us ha he e idence ga he ed om he p esen doc o al wo k
can be use ul o u u e heo e ical p oposals ha ake in o conside a ion he mul ile el
s uc u e (i.e., neu ocogni i e, beha io al, con ex ual) o language de elopmen . We
hope ha ou indings can in o m eaching and clinical p ac ices o child en g owing
in bilingual en i onmen s, by highligh ing speci ic en i onmen al ac o s ha , like he
amoun o linguis ic exposu e, shape language knowledge and p o iciency o a g ea
ex en and should be aken in o accoun when assessing ypical and a ypical language
de elopmen .
117
Re e ences
Acha, J., Laka, I., Landa, J., & Salabu u, P. (2014). EHME: A New Wo d Da abase o
Resea ch in Basque Language. The Spanish Jou nal o Psychology, 17.
h ps://doi.o g/10.1017/sjp.2014.79
Ahissa , M., Lubin, Y., Pu e -Ka z, H., & Banai, K. (2006). Dyslexia and he ailu e o
o m a pe cep ual ancho . Na u e Neu oscience, 9(12), A icle 12.
h ps://doi.o g/10.1038/nn1800
Alexand ou, A. M., Saa inen, T., Kujala, J., & Salmelin, R. (2018). Co ical en ainmen :
Wha we can lea n om s udying na u alis ic speech pe cep ion. Language,
Cogni ion and Neu oscience, 0(0), 1–13.
h ps://doi.o g/10.1080/23273798.2018.1518534
Al man, C., Golds ein, T., & A mon-Lo em, S. (2018). Vocabula y, me alinguis ic
awa eness and language dominance among bilingual p eschool child en.
F on ie s in Psychology, 9. h ps://doi.o g/10.3389/ psyg.2018.01953
Ande son, R. M., Giezen, M. R., & Pou quié, M. (2018). Basque-Spanish bilingual
child en’s exp essi e and ecep i e g amma ical abili ies. Linguis ic App oaches
o Bilingualism. h ps://doi.o g/10.1075/lab.17034.and
An hony, J. L., & Lonigan, C. J. (2004). The Na u e o Phonological Awa eness:
Con e ging E idence F om Fou S udies o P eschool and Ea ly G ade School
Child en. Jou nal o Educa ional Psychology, 96(1), 43–55.
h ps://doi.o g/10.1037/0022-0663.96.1.43
118
An hony, J. L., Lonigan, C. J., D iscoll, K., Phillips, B. M., & Bu gess, S. R. (2003).
Phonological sensi i i y: A quasi-pa allel p og ession o wo d s uc u e uni s and
cogni i e ope a ions. Reading Resea ch Qua e ly, 38(4), 470–487.
h ps://doi.o g/10.1598/RRQ.38.4.3
A ani i, A. (2009). Rhy hm, Timing and he Timing o Rhy hm. Phone ica, 66(1–2), 46–
63. h ps://doi.o g/10.1159/000208930
A ahe i, A., Choisdealbha, Á. N., Di Libe o, G. M., Rocha, S., B usini, P., Mead, N.,
Olawole-Sco , H., Bou is, P., Gibbon, S., Williams, I., G ey, C., Flanagan, S., &
Goswami, U. (2022). Del a- and he a-band co ical acking and phase-
ampli ude coupling o sung speech by in an s. Neu oImage, 247, 118698.
h ps://doi.o g/10.1016/j.neu oimage.2021.118698
A ahe i, A., Choisdealbha, Á. N., Libe o, G. M. D., Rocha, S., B usini, P., Mead, N.,
Olawole-Sco , H., Bou is, P., Gibbon, S., Williams, I., G ey, C., Flanagan, S., &
Goswami, U. (2021). Del a- and he a-band co ical acking and phase-ampli ude
coupling o sung speech by in an s (p. 2020.10.12.329326).
h ps://doi.o g/10.1101/2020.10.12.329326
A ig, M., & Weine , S. (2020). Wha Impac s Ea ly Language Skills? E ec s o Social
Dispa i ies and Di e en P ocess Cha ac e is ics o he Home Lea ning
En i onmen in he Fi s 2 Yea s. F on ie s in Psychology, 11.
h ps://www. on ie sin.o g/a icle/10.3389/ psyg.2020.557751
Banai, K., & Ahissa , M. (2018). Poo sensi i i y o sound s a is ics impai s he
acquisi ion o speech ca ego ies in dyslexia. Language, Cogni ion and
Neu oscience, 33(3), 321–332. h ps://doi.o g/10.1080/23273798.2017.1408851
125
Jou nal o Speech, Language, and Hea ing Resea ch, 63(7), 2453–2467.
h ps://doi.o g/10.1044/2020_JSLHR-19-00229
Ding, N., Melloni, L., Yang, A., Wang, Y., Zhang, W., & Poeppel, D. (2017).
Cha ac e izing Neu al En ainmen o Hie a chical Linguis ic Uni s using
Elec oencephalog aphy (EEG). F on ie s in Human Neu oscience, 11, 481.
h ps://doi.o g/10.3389/ nhum.2017.00481
Ding, N., Pa el, A. D., Chen, L., Bu le , H., Luo, C., & Poeppel, D. (2017). Tempo al
modula ions in speech and music. Neu oscience & Biobeha io al Re iews, 81, 181–
187. h ps://doi.o g/10.1016/j.neubio e .2017.02.011
Ding, N., & Simon, J. Z. (2012). Neu al coding o con inuous speech in audi o y co ex
du ing monau al and dicho ic lis ening. Jou nal o Neu ophysiology, 107(1), 78–
89. h ps://doi.o g/10.1152/jn.00297.2011
Doelling, K. B., A nal, L. H., Ghi za, O., & Poeppel, D. (2014). Acous ic landma ks d i e
del a- he a oscilla ions o enable speech comp ehension by acili a ing
pe cep ual pa sing. Neu oImage, 85 P 2, 761–768.
h ps://doi.o g/10.1016/j.neu oimage.2013.06.035
Döpke, S. (1998). Compe ing language s uc u es: The acquisi ion o e b placemen by
bilingual Ge man-English child en. Jou nal o Child Language, 25(3), 555–584.
h ps://doi.o g/10.1017/S0305000998003584
Duchon, A., Pe ea, M., Sebas ián-Gallés, N., Ma í, A., & Ca ei as, M. (2013). EsPal:
One-s op shopping o Spanish wo d p ope ies. Beha io Resea ch Me hods,
45(4), 1246–1258. h ps://doi.o g/10.3758/s13428-013-0326-1

126
Ezeizaba ena, M.-J., & Ga cía, A. (2015). Ea ly coda p oduc ion in bilingual Spanish and
Basque. In The Acquisi ion o Spanish in Unde s udied Language Pai ings. Edi ed
by Ti any Judy & Sil ia Pe piñán. John Benjamins Publishing Company.
Ezeizaba ena, M.-J., & Ga cia Fe nandez, I. (2018). Leng h o U e ance, in Mo phemes
o in Wo ds?: MLU3-w, a Reliable Measu e o Language De elopmen in Ea ly
Basque. F on ie s in Psychology, 8.
h ps://www. on ie sin.o g/a icles/10.3389/ psyg.2017.02265
Fe nald, A. (1985). Fou -mon h-old in an s p e e o lis en o mo he ese. In an Beha io
and De elopmen , 8(2), 181–195. h ps://doi.o g/10.1016/S0163-6383(85)80005-9
Fe nald, A. (2000). Speech o In an s as Hype speech: Knowledge-D i en P ocesses in
Ea ly Wo d Recogni ion. Phone ica, 57(2–4), 242–254.
h ps://doi.o g/10.1159/000028477
Fe nald, A., & Simon, T. (1984). Expanded in ona ion con ou s in mo he s’ speech o
newbo ns. De elopmen al Psychology, 20(1), 104–113.
h ps://doi.o g/10.1037/0012-1649.20.1.104
Fe nald, A., Taeschne , T., Dunn, J., Papousek, M., Boysson-Ba dies, B. de, & Fukui, I.
(1989). A c oss-language s udy o p osodic modi ica ions in mo he s’ and a he s’
speech o p e e bal in an s*. Jou nal o Child Language, 16(3), 477–501.
h ps://doi.o g/10.1017/S0305000900010679
Foulke, E., & S ich , T. G. (1969). Re iew o esea ch on he in elligibili y and
comp ehension o accele a ed speech. Psychological Bulle in, 72(1), 50–62.
h ps://doi.o g/10.1037/h0027575
127
F ase , J., Goswami, U., & Con i-Ramsden, G. (2010). Dyslexia and Speci ic Language
Impai men : The Role o Phonology and Audi o y P ocessing. Scien i ic S udies
o Reading, 14(1), 8–29. h ps://doi.o g/10.1080/10888430903242068
F izelle, P., Thompson, P. A., Mcdonald, D., & Bishop, D. V. M. (2018). G ow h in
syn ac ic complexi y be ween ou yea s and adul hood: E idence om a
na a i e ask. Jou nal o Child Language, 45(5), 1174–1197.
h ps://doi.o g/10.1017/S0305000918000144
Fu ow, D., Nelson, K., & Benedic , H. (1979). Mo he s’ speech o child en and syn ac ic
de elopmen : Some simple ela ionships*. Jou nal o Child Language, 6(3), 423–
442. h ps://doi.o g/10.1017/S0305000900002464
Gámez, P. B., G iskell, H. L., Sob e illa, Y. N., & Vazquez, M. (2019). Dual Language and
English-Only Lea ne s’ Exp essi e and Recep i e Language Skills and Exposu e
o Pee s’ Language. Child De elopmen , 90(2), 471–479.
h ps://doi.o g/10.1111/cde .13197
Ga cia-Sie a, A., Ramí ez-Espa za, N., & Kuhl, P. K. (2016). Rela ionships be ween
quan i y o language inpu and b ain esponses in bilingual and monolingual
in an s. In e na ional Jou nal o Psychophysiology, 110, 1–17.
h ps://doi.o g/10.1016/j.ijpsycho.2016.10.004
Ga he cole, S. E. (2006). Nonwo d epe i ion and wo d lea ning: The na u e o he
ela ionship. Applied Psycholinguis ics, 27(4), 513–543.
h ps://doi.o g/10.1017/S0142716406060383
Ga he cole, S. E., & Baddeley, A. D. (1989). E alua ion o he ole o phonological STM
in he de elopmen o ocabula y in child en: A longi udinal s udy. Jou nal o
128
Memo y and Language, 28(2), 200–213. h ps://doi.o g/10.1016/0749-
596X(89)90044-2
Ga he cole, S. E., Willis, C., Emslie, H., & Baddeley, A. D. (1991). The in luences o
numbe o syllables and wo dlikeness on child en’s epe i ion o nonwo ds.
Applied Psycholinguis ics, 12(3), 349–367.
h ps://doi.o g/10.1017/S0142716400009267
Ga he cole, S. E., Willis, C. S., Emslie, H., & Baddeley, A. D. (1992). Phonological
memo y and ocabula y de elopmen du ing he ea ly school yea s: A
longi udinal s udy. De elopmen al Psychology, 28, 887–898.
h ps://doi.o g/10.1037/0012-1649.28.5.887
Ga he cole, V. C. M., & Thomas, E. M. (2005). Mino i y language su i al: Inpu ac o s
in luencing he acquisi ion o Welsh. ISB4. P oceedings o he 4 h In e na ional
Symposium on Bilingualism, Some ille, MA.
Ge ain, J., & We ke , J. F. (2013). P osody cues wo d o de in 7-mon h-old bilingual
in an s. Na u e Communica ions, 4, 1490. h ps://doi.o g/10.1038/ncomms2430
Ghi za, O. (2011). Linking Speech Pe cep ion and Neu ophysiology: Speech Decoding
Guided by Cascaded Oscilla o s Locked o he Inpu Rhy hm. F on ie s in
Psychology, 2. h ps://doi.o g/10.3389/ psyg.2011.00130
Ghi za, O. (2012). On he Role o The a-D i en Syllabic Pa sing in Decoding Speech:
In elligibili y o Speech wi h a Manipula ed Modula ion Spec um. F on ie s in
Psychology, 3. h ps://doi.o g/10.3389/ psyg.2012.00238
129
Ghi za, O. (2017). Acous ic-d i en del a hy hms as p osodic ma ke s. Language,
Cogni ion and Neu oscience, 32(5), 545–561.
h ps://doi.o g/10.1080/23273798.2016.1232419
Gibson, T. A., Olle , D. K., Ja mulowicz, L., & E hing on, C. A. (2012). The ecep i e–
exp essi e gap in he ocabula y o young second-language lea ne s: Robus ness
and possible mechanisms*. Bilingualism: Language and Cogni ion, 15(1), 102–116.
h ps://doi.o g/10.1017/S1366728910000490
Gigue e, D., & Ho , E. (2022). Bilingual de elopmen in he ecep i e and exp essi e
domains: They di e . In e na ional Jou nal o Bilingual Educa ion and
Bilingualism, 0(0), 1–10. h ps://doi.o g/10.1080/13670050.2022.2087039
Gi aud, A.-L., & Poeppel, D. (2012). Co ical oscilla ions and speech p ocessing:
Eme ging compu a ional p inciples and ope a ions. Na u e Neu oscience, 15(4),
511–517. h ps://doi.o g/10.1038/nn.3063
Gooch, D., & Snowling, M. (2018). De elopmen al Ou comes o Child en a High Risk
o Dyslexia and Child en wi h De elopmen al Language Diso de . Child
De elopmen . h ps://onlinelib a y.wiley.com/jou nal/14678624
Goswami, U. (2011). A empo al sampling amewo k o de elopmen al dyslexia. T ends
in Cogni i e Sciences, 15(1), 3–10. h ps://doi.o g/10.1016/j. ics.2010.10.001
Goswami, U. (2017). A Neu al Basis o Phonological Awa eness? An Oscilla o y
Tempo al-Sampling Pe spec i e. Cu en Di ec ions in Psychological Science, 56–
63. h ps://doi.o g/10.1177/0963721417727520
130
Goswami, U., & B yan , P. (1990). Phonological skills and lea ning o ead (pp. iii, 166).
Law ence E lbaum Associa es, Inc.
Goswami, U., Cumming, R., Chai , M., Huss, M., Mead, N., Wilson, A. M., Ba nes, L., &
Foske , T. (2016). Pe cep ion o Fil e ed Speech by Child en wi h De elopmen al
Dyslexia and Child en wi h Speci ic Language Impai men s. F on ie s in
Psychology, 7. h ps://doi.o g/10.3389/ psyg.2016.00791
Goswami, U., Ge son, D., & As uc, L. (2010). Ampli ude en elope pe cep ion,
phonology and p osodic sensi i i y in child en wi h de elopmen al dyslexia.
Reading and W i ing, 23(8), 995–1019. h ps://doi.o g/10.1007/s11145-009-9186-
6
Goswami, U., & Leong, V. (2013). Speech hy hm and empo al s uc u e: Con e ging
pe spec i es? Labo a o y Phonology, 4(1). h ps://doi.o g/10.1515/lp-2013-0004
Goswami, U., Thomson, J., Richa dson, U., S ain ho p, R., Hughes, D., Rosen, S., & Sco ,
S. K. (2002). Ampli ude en elope onse s and de elopmen al dyslexia: A new
hypo hesis. P oceedings o he Na ional Academy o Sciences, 99(16), 10911–10916.
h ps://doi.o g/10.1073/pnas.122368599
Goswami, U., Wang, H.-L. S., C uz, A., Foske , T., Mead, N., & Huss, M. (2010).
Language-uni e sal Senso y De ici s in De elopmen al Dyslexia: English,
Spanish, and Chinese. Jou nal o Cogni i e Neu oscience, 23(2), 325–337.
h ps://doi.o g/10.1162/jocn.2010.21453
G anados Ba be o, R., Ghesquiè e, P., & Wou e s, J. (2022). De elopmen o A ypical
Reading a Ages 5 o 9 Yea s and P ocessing o Speech En elope Modula ions in

131
he B ain. F on ie s in Compu a ional Neu oscience, 16.
h ps://www. on ie sin.o g/a icles/10.3389/ ncom.2022.894578
G eenbe g, S. (2006). A Mul i-Tie F amewo k o Unde s anding Spoken Language. In
Lis ening o speech: An audi o y pe spec i e (pp. 411–433). Law ence E lbaum
Associa es Publishe s.
G eenbe g, S., Ca ey, H., Hi chcock, L., & Chang, S. (2003). Tempo al p ope ies o
spon aneous speech—A syllable-cen ic pe spec i e. Jou nal o Phone ics, 31(3),
465–485. h ps://doi.o g/10.1016/j.wocn.2003.09.005
G oss, J., Hoogenboom, N., Thu , G., Schyns, P., Panze i, S., Belin, P., & Ga od, S. (2013).
Speech Rhy hms and Mul iplexed Oscilla o y Senso y Coding in he Human
B ain. PLOS Biology, 11(12), e1001752.
h ps://doi.o g/10.1371/jou nal.pbio.1001752
Gulli e , J. W., Chai, X. J., Whi o d, V., Pi ne a, I., Baum, S., Klein, D., & Ti one, D.
(2018). Bilingual expe ience and es ing-s a e b ain connec i i y: Impac s o L2
age o acquisi ion and social di e si y o language use on con ol ne wo ks.
Neu opsychologia, 117, 123–134.
h ps://doi.o g/10.1016/j.neu opsychologia.2018.04.037
Gulli e , J. W., Kousaie, S., Gilbe , A. C., G an , A. M., Gi oud, N., Coul e , K., Klein, D.,
Baum, S., Phillips, N., & Ti one, D. (2020). Bilingual language expe ience as a
mul idimensional spec um: Associa ions wi h objec i e and subjec i e language
p o iciency [P ep in ]. PsyA Xi . h ps://doi.o g/10.31234/os .io/gb9nd
132
Gulli e , J. W., & Ti one, D. (2019). Cha ac e izing he social di e si y o bilingualism
using language en opy. Bilingualism: Language and Cogni ion, No Pagina ion
Speci ied-No Pagina ion Speci ied. h ps://doi.o g/10.1017/S1366728919000026
Guo, L.-Y., & Eisenbe g, S. (2015). Sample Leng h A ec s he Reliabili y o Language
Sample Measu es in 3-Yea -Olds: E idence F om Pa en -Elici ed Con e sa ional
Samples. Language, Speech, and Hea ing Se ices in Schools, 46(2), 141–153.
h ps://doi.o g/10.1044/2015_LSHSS-14-0052
Guo, L.-Y., Schneide , P., & Ha ison, W. (2021). Clausal Densi y Be ween Ages 4 and 9
Yea s o he Edmon on Na a i e No ms Ins umen : Re e ence Da a and
Psychome ic P ope ies. Language, Speech, and Hea ing Se ices in Schools,
52(1), 354–368. h ps://doi.o g/10.1044/2020_LSHSS-20-00043
Gu ié ez-Clellen, V. F., & Simon-Ce eijido, G. (2010). Using Nonwo d Repe i ion Tasks
o he Iden i ica ion o Language Impai men in Spanish-English-Speaking
Child en: Does he Language o Assessmen Ma e ? Lea ning Disabili ies
Resea ch & P ac ice, 25(1), 48–58. h ps://doi.o g/10.1111/j.1540-
5826.2009.00300.x
Haake, M., Hansson, K., Gulz, A., Schö z, S., & Sahlén, B. (2014). The slowe he be e ?
Does he speake ’s speech a e in luence child en’s pe o mance on a language
comp ehension es ? In e na ional Jou nal o Speech-Language Pa hology, 16(2),
181–190. h ps://doi.o g/10.3109/17549507.2013.845690
Hämäläinen, J. A., Salminen, H. K., & Leppänen, P. H. T. (2012). Basic Audi o y
P ocessing De ici s in Dyslexia: Sys ema ic Re iew o he Beha io al and E en -
133
Rela ed Po en ial/ Field E idence. Jou nal o Lea ning Disabili ies, 46(5), 413–
427. h ps://doi.o g/10.1177/0022219411436213
Haman, E., Wodniecka, Z., Ma ecka, M., Szewczyk, J., Białecka-Pikul, M., O winowska,
A., Mieszkowska, K., Łuniewska, M., Kołak, J., Miękisz, A., Kacp zak, A., Banasik,
N., & Fo yś-Nogala, M. (2017). How Does L1 and L2 Exposu e Impac L1
Pe o mance in Bilingual Child en? E idence om Polish-English Mig an s o
he Uni ed Kingdom. F on ie s in Psychology, 8.
h ps://www. on ie sin.o g/a icles/10.3389/ psyg.2017.01444
Hamil on, L. S., Edwa ds, E., & Chang, E. F. (2018). A Spa ial Map o Onse and
Sus ained Responses o Speech in he Human Supe io Tempo al Gy us. Cu en
Biology, 28(12), 1860-1871.e4. h ps://doi.o g/10.1016/j.cub.2018.04.033
Hamil on, L. S., & Hu h, A. G. (2020). The e olu ion will no be con olled: Na u al
s imuli in speech neu oscience. Language, Cogni ion and Neu oscience, 35(5),
573–582. h ps://doi.o g/10.1080/23273798.2018.1499946
Ha , B., & Risley, T. R. (1995). Meaning ul di e ences in he e e yday expe ience o young
Ame ican child en (pp. xxiii, 268). Paul H B ookes Publishing.
Hauswald, A., Kei el, A., Chen, Y.-P., Rösch, S., & Weisz, N. (2022). Deg ada ion le els
o con inuous speech a ec neu al speech acking and alpha powe di e en ly.
Eu opean Jou nal o Neu oscience, 55(11–12), 3288–3302.
h ps://doi.o g/10.1111/ejn.14912
Hi ose, K., & Kawanami, H. (2002). Tempo al a e change o dialogue speech in
p osodic uni s as compa ed o ead speech. Speech Communica ion, 36(1), 97–111.
h ps://doi.o g/10.1016/S0167-6393(01)00028-0
134
Ho , E. (2003a). Causes and consequences o SES- ela ed di e ences in pa en - o-child
speech. In Socioeconomic s a us, pa en ing, and child de elopmen (pp. 147–160).
Law ence E lbaum Associa es Publishe s.
Ho , E. (2003b). The Speci ici y o En i onmen al In luence: Socioeconomic S a us
A ec s Ea ly Vocabula y De elopmen Via Ma e nal Speech. Child De elopmen ,
74(5), 1368–1378. h ps://doi.o g/10.1111/1467-8624.00612
Ho , E., Co e, C., Place, S., Rumiche, R., Seño , M., & Pa a, M. (2012). Dual language
exposu e and ea ly bilingual de elopmen . Jou nal o Child Language, 39(1), 1–27.
h ps://doi.o g/10.1017/S0305000910000759
Ho -Ginsbe g, E. (1986). Func ion and s uc u e in ma e nal speech: Thei ela ion o
he child’s de elopmen o syn ax. De elopmen al Psychology, 22(2), 155–163.
h ps://doi.o g/10.1037/0012-1649.22.2.155
Hosoda, C., Tanaka, K., Na iai, T., Honda, M., & Hanakawa, T. (2013). Dynamic Neu al
Ne wo k Reo ganiza ion Associa ed wi h Second Language Vocabula y
Acquisi ion: A Mul imodal Imaging S udy. Jou nal o Neu oscience, 33(34),
13663–13672. h ps://doi.o g/10.1523/JNEUROSCI.0410-13.2013
Howell, P., & Kadi-Hani i, K. (1991). Compa ison o p osodic p ope ies be ween ead
and spon aneous speech ma e ial. Speech Communica ion, 10(2), 163–169.
h ps://doi.o g/10.1016/0167-6393(91)90039-V
Huang, B. H., Chang, Y.-H. S., Zhi, M., & Niu, L. (2020). The e ec o inpu on bilingual
adolescen s’ long- e m language ou comes in a o eign language ins uc ion
con ex . In e na ional Jou nal o Bilingualism, 24(1), 8–25.
h ps://doi.o g/10.1177/1367006918768311