Discrimination of Second Language Vowel Contrasts and the Role of Phonological Short-Term Memory and Nonverbal Intelligence

Abstract

Although extensive research has focused on the perceptual abilities of second language (L2) learners, a significant gap persists in understanding how cognitive functions like phonological short-term memory (PSTM) and nonverbal intelligence (IQ) impact L2 speech perception. This study sets out to investigate the discrimination of L2 English monophthongal vowel contrasts and to assess the effect of PSTM and nonverbal IQ on L2 speech perception. The participants consisted of adult monolingually-raised Greek speakers, who completed an AX discrimination test, a digit span test, and a nonverbal intelligence test. A control group of English speakers also completed the AX test. Data were analyzed using Bayesian regression models. The results revealed that Greek speakers exhibited below chance discrimination for the majority of L2 vowel contrasts, consistently underperforming in comparison to the control group. Intriguingly, the study did not provide substantial evidence in favor of more accurate discrimination of L2 contrasts by Greek participants with high PSTM compared to those with low PSTM. However, the study yielded compelling evidence indicating that Greek participants with higher IQ demonstrated superior accuracy in discriminating most L2 contrasts compared to their lower IQ counterparts. The limited influence of PSTM on speech perception suggests the need for further exploration, considering the potential impact of test methodologies and the intricate interplay of other confounding factors. Furthermore, the study uncovers a noteworthy relationship between nonverbal IQ and L2 speech perception, likely linked with the association of high IQ with enhanced attentional capacities, information processing abilities, and learning skills—all of which are pivotal for accurate speech perception.

Full text

Vol.:(0123456789)
Journal of Psycholinguistic Research (2024) 53:9
https://doi.org/10.1007/s10936-024-10038-z
1 3
Discrimination ofSecond Language Vowel Contrasts
andtheRole ofPhonological Short‑Term Memory
andNonverbal Intelligence
GeorgiosP.Georgiou1,2 · AretousaGiannakou1,2
Accepted: 4 December 2023 / Published online: 4 February 2024
© The Author(s) 2024
Abstract
Although extensive research has focused on the perceptual abilities of second language
(L2) learners, a significant gap persists in understanding how cognitive functions like pho-
nological short-term memory (PSTM) and nonverbal intelligence (IQ) impact L2 speech
perception. This study sets out to investigate the discrimination of L2 English monoph-
thongal vowelcontrasts and to assess the effect of PSTM and nonverbalIQ on L2 speech
perception. The participants consisted of adult monolingually-raised Greek speakers, who
completed an AX discrimination test, a digit span test, and a nonverbal intelligence test.
A control group of English speakers also completed the AX test. Data were analyzed
using Bayesian regression models. The results revealed that Greek speakers exhibited
belowchance discrimination for the majority of L2 vowel contrasts, consistently underper-
forming in comparison to the control group. Intriguingly, the study did not provide substan-
tial evidence in favor of more accurate discrimination of L2 contrasts by Greekparticipants
with high PSTM compared to those with low PSTM. However, the study yielded com-
pelling evidence indicating that Greekparticipants with higher IQ demonstrated superior
accuracy in discriminating most L2 contrasts compared to their lowerIQ counterparts. The
limited influence of PSTM on speech perception suggests the need for further exploration,
considering the potential impact of test methodologies and the intricate interplay of other
confounding factors. Furthermore, the study uncovers a noteworthy relationship between
nonverbal IQ and L2 speech perception, likely linked with the association of high IQ with
enhanced attentional capacities, information processing abilities, and learning skills—all
of which are pivotal for accurate speech perception.
Keywords Speech perception· Phonological short-term memory· Nonverbal intelligence·
Second language· Greek· English
* Georgios P. Georgiou
georgiou.georg@unic.ac.cy
Aretousa Giannakou
giannakou.a@unic.ac.cy
1 Department ofLanguages andLiterature, University ofNicosia, Nicosia, Cyprus
2 Phonetic Lab, University ofNicosia, Nicosia, Cyprus
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 2 of 18
Introduction
The difficulties observed in adults in accurately distinguishing second language (L2) sound
contrasts is mainly attributed to the strong effect of their first language (L1) (Evans &
Alshangiti, 2018; Georgiou, 2022a, 2024; Iverson etal., 2003; Shinohara & Iverson, 2018).
Considering that L2 sounds are mapped in terms of L1 categories in the human mind (e.g.,
see Best & Tyler, 2007; Cebrian, 2006; Georgiou etal., 2020a, 2020b; Mayr & Escudero,
2010; Schmidt, 2018), L2 sounds mapped to a single L1 category are particularly difficult
to distinguish. For example, Georgiou etal. (2020b) found that Russian learners of English
struggled to discriminate the English /æ/–/e/contrast since they assimilated both English
vowels to the same Russian category, that is, /e/. This single-category mapping might be
fostered by the size and complexity of the L1-L2 systems. For instance, Iverson and Evans
(2007) reported that speakers of L1s with large and more complex vowel systems (German
and Norwegian) identified the L2 English vowels more accurately than speakers of L1s
with smaller and less complex vowel systems (Spanish and French). Nevertheless, more
recent studies argued that having a large and more complex vowel system does not guar-
antee a discrimination advantage. Elvin etal. (2014) found that Australian English listen-
ers do not discriminate Brazilian Portuguese vowel contrasts better than Spanish listeners
despite the large size and complexity of the Australian English vowel system and the small
size and simplicity of the Spanish vowel system. The authors concluded that crosslinguis-
tic properties but not crosslinguistic vowel systems predicted the difficulty of learners in
discriminating the L2 contrasts. The size and complexity of L1 and L2 systems can be con-
sidered together with the acoustic similarity between L1-L2 sounds to predict the speakers’
L2 perception patterns (see Alispahic etal., 2017; Georgiou, 2023d).
Discrimination predictions are usually based on the theoretical account of specific
speech models. A recent addition is the Universal Perceptual Model (UPM) (Georgiou &
Dimitriou, 2023; Georgiou, 2021a, 2023c), which accounts for the difficulties of learners
in the discrimination of L2 sound contrasts. Being consistent with other models, UPM also
suggests that the acoustic characteristics of L1-L2 sounds may be useful for the assessment
of speakers’ discrimination abilities over the L2 contrasts. More specifically, it proposes
that perceptual overlapping between two contrastive L2sounds against an L1 sound can
predict the discrimination difficulty of the L2 pair. In this context, two L2 contrasts can
be completely overlapping, partially overlapping and nonoverlapping. Completely overlap-
ping contrasts are expected to be discriminated worse than partially overlapping contrasts,
and the latter are expected to be discriminated worse than nonoverlapping contrasts.
Apart from the linguistic account, the contribution of several factors such as age of
learning (Flege etal., 1995), length of residence (Meador et al., 2000), L1-L2 use (Flege
et al., 1997), quality and quantity of L2 input (Flege & Liu, 2001), gender (Flege &
Fletcher, 1992), etc. has been investigated in a large number of studies. The role of cogni-
tive functions such as phonological short-term memory (PSTM) and intelligence (IQ) has
received less scientific attention. PSTM constitutes an integral component of the multidi-
mensional working memory (WM) system, alongside dimensions associated with atten-
tional resource capacity and allocation, as well as processing speed (Montgomery etal.,
2009). Baddeley etal. (1984) reported that the PSTM model contains two main elements: a
phonological buffer or store designed for the temporary retention of memory traces lasting
a few seconds, and a subvocal rehearsal process employed for the purpose of rejuvenating
these memory traces. General intelligence, or the so-called g factor, originally proposed by
Spearman (1904), comprises a psychometric construction of cognitive abilities and human
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
Page 3 of 18 9
intelligence. Cattell (1963) refers to two types of general intelligence, fluid and crystal-
lized. Fluid intelligence shows the ability of individuals to solve problems and reason and
is often associated with nonverbal skills. Crystallized intelligence emerges from past expe-
rience and is improved over time as individuals increase their knowledge; this knowledge
is language- and culture-specific (Dryden etal., 2017). This type of intelligence is based on
verbal skills.
Previous work that attempted to define the relationship between PSTM and language
studied the effect of PSTM on L2 oral fluency (O’Brien etal., 2006), vocabulary learning
(French & O’Brien, 2008; Martin & Ellis, 2012), and grammar aspects (Masoura & Gath-
erole, 2005). There is lessevidence regarding the interface betweenPSTM and L2 speech
perception. A relationship between PSTM and speech perception patterns is expected
since PSTM enables learners to temporarily store nonnative sounds, facilitating the abil-
ity to perceive and differentiate them from the familiar sounds of their L1. By employing
the rehearsal mechanism, nonnative sounds can effectively be reinforced, thus aiding in
the development of enduring and robust mental representations. This, in turn, enhances
the capacity for long-term recognition and differentiation of these sounds. MacKay etal.
(2001) delved into the identification of English vowels by Italian speakers. Their research
unveiled that PSTM significantly predicted the identification of word-final English conso-
nants, accounting for approximately 15% of the variance, as well as the identification of
word-initial English consonants, where it explained about 8% of the variance. These find-
ings were deemed noteworthy for understanding and predicting the identification patterns
exhibited by the speakers. Lengeris and Nicolaides (2014) examined the identification of
English consonants by Greek listeners in quiet and noise contextsand the effect of PSTM
on this identification using a nonword repetition task. The findings showed that PSTM
highly correlated with the capacity of the listeners to identify English consonants. Aliaga-
Garcia etal. (2011) indicated that Catalan-Spanish learners of English with higher PSTM
capacities had higher accuracy scores and perceptual gains from phonetic training com-
pared to Catalan-Spanish learners of English with lower PSTM capacities, highlighting
the significant role of PSTM in L2 vowel discrimination. Along the same lines, Safronova
(2016) found a positive correlation between PSTM and discrimination performance over
L2 English contrasts by Spanish-Catalan learners of English. In addition, larger PSTM
capacity was associated with better performance in the distinction of these contrasts. How-
ever, this result could not be verified for Azerbaijani learners of English (Ghaffarvand
Mokari & Werner, 2019). The authors trained the learners in the discrimination of L2 Eng-
lish contrasts, concluding that there was no correlation between gains from high variability
phonetic training and PSTM—thus, the role of PSTM in phonological learning was insig-
nificant. Similarly, Safronova and Mora (2012) indicated no advantage for Catalan learn-
ers of English with a large PSTM capacity in more accurately identifying the English /
iː/–/ɪ/ contrast in comparison to learners with a low PSTM capacity. PSTM is measured
using a variety of methods including nonword repetition (Lengeris & Nicolaides, 2014) or
serial nonword recognition tasks (SNWR) (Cerviño-Povedano & Mora, 2010; Ghaffarvand
Mokari & Werner, 2019), immediate serial recall (Tree & Playfoot, 2019) and digit span
tests (Kim etal., 2016) among other. Although the list of studies is not exhaustive, previ-
ous findings suggest mixed results regarding the effect of PSTM on the learners’ speech
perception abilities.
There is evidence focusing on the link between L2 learning and IQ. In some studies,
high- vs low-achieving L2 learners were found to differ from each other in terms of verbal/
nonverbalIQ (e.g., Sparks etal., 1992), while some other studies indicated no such differ-
ence (e.g., Sparks etal., 2012). In an earlier study, Carroll (1962)suggested that while L2
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 4 of 18
aptitude does not depend on IQ, there might be an association between generalIQ and the
capability of cognitive mechanisms insupporting L2 proficiency. Łockiewicz etal. (2018)
explored the predictors of foreign language learning by L1 Polish preschooler learners of
Englishas a foreign language. The authors observed that nonverbal IQ was a good predic-
tor of the oral English skills of children. Woumans etal. (2019) examined the relationship
between L2 acquisition and nonverbal cognitive abilities, identifyingnonverbalIQ as one
of the contributing factors to the acquisition of L2 Dutch vocabulary, as higher perfor-
mance on Dutch L2 vocabulary correlated with higher nonverbalIQ progress. A similar
positive relationship between nonverbalIQ and L2 vocabulary development was observed
by Daller and Ongun (2018) in Turkish-English successive bilingual children. Research
regarding the effect of IQ on phonetic aspectsis scarce.Rota and Reiterer (2009) examined
among other the correlation between verbal and nonverbal IQ and phonetic abilities. Their
results suggested no correlation between verbal and nonverbal IQ and pronunciation, while
a correlation was found between verbal IQ and phonetic coding ability. Similarly, Chris-
tiner etal. (2018) reported that nonverbalIQ did not correlate with language phonetic apti-
tude. In contrast, one recent study demonstrated that Cypriot Greek speakers of L2 English
with a high nonverbal IQ capacity discriminated the majority of L2 vowel contrasts more
accurately than the corresponding speakers with a low nonverbal IQ capacity (Georgiou,
2023a). Some of the most popular measures of nonverbal IQ are the Wechsler Abbrevi-
ated Scale of Intelligence (WASI) (Wechsler, 1999), the Wechsler Adult Intelligence Scale
(WAIS) (Wechsler, 2008), and the Raven Progressive Matrices tests (Raven etal., 2000).
All in all, the very few studies in the literature manifest contradictory findings about the
effect of nonverbalIQ on L2 speech perception.
This paper aims to answer two main questions: (i) how do Greek speakers of L2 Eng-
lish discriminate specific pairs of English monophthongal vowel contrasts (/iː/–/ɪ/, /iː/–/e/,
/ɑː/–/ʌ/, /æ/–/ɑː/, /ɔː/–/ɒ/, and /uː/–/ʊ/)? and (ii) how do the Greekspeakers’ PSTM and
nonverbalIQ capacities affect their discrimination abilities? The first question’s motiva-
tion stems from the limited research on the discrimination of English sound contrasts by
Greek speakers. This research gap offers an opportunity to gain valuable insights into how
an individual’s L1 influences the acquisition of L2 sounds, especially in the case of an
understudied language like Greek. By exploring this, we can enhance our understanding of
crosslinguistic phonology and, practically, help educators develop effective teaching strate-
gies and providelanguage learners with tools to identify phoneticchallenges. The second
question arises from the noticeable scarcity of comprehensive research into the influence of
PSTM and nonverbalIQ on L2 speech perception, especially in speakers with a Greek L1
background. This gap in knowledge hinders our ability to grasp why some language learn-
ers thrive in acquiring new phonological and phonetic elements while others encounter dif-
ficulties. Exploring the connection between PSTM, nonverbalIQ, and L2 speech percep-
tion can illuminate this phenomenon, contributing to a more profound comprehension of
the cognitive underpinnings of language acquisition. This understanding can be a pivotal
resource for educators seeking to tailor instruction to the cognitive profiles of individual
learners, ultimately enhancing language learning outcomes.
Standard Modern Greek consists of a simple five-vowel system including vowels/i e
a o u/ (see Georgiou & Themistocleous, 2021), while Standard Southern British English
has a more extensive vowel system which includes tense and lax vowels, namely, /iː uː ɜː
ɔː ɑː/ and /ɪ ʊ e æ ʌ ɒ/ respectively. This study concentrates on monophthongs as the few
previous studies involving Greek speakers have focused on this type of L2 English vowels.
As shown by previous research,most of these sounds are difficult for Greek speakers due
to the size and complexity of the L2 vowel system and the differences between the L1 and
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
Page 5 of 18 9
the L2 vowels at the acousticophonetic level. For example, Lengeris (2009) studied the
perception of English vowels by adult Greek learners of English, indicating sinlge-category
assimilations for several English vowels. Specifically,English /iː/ and /ɪ/ were both assimi-
lated to Greek category /i/, English /ε/ and /ɜː/ to Greek /e/, English /æ/ and /ʌ/ to Greek
/a/, English /ɑː/, /ɒ/ and /ɔː/ to Greek /o/ and English /ʊ/ and /u:/ to Greek /u/. The dis-
crimination scores showed that the most difficult contrast was /uː/–/ʊ/ followed by /ɑː/–/ʌ/,
/ɔː/–/uː/, /æ/– /ʌ/, /iː/–/ɪ/, /ɔː/– /ɑː/, /ɒ/ –/ɔː/, /æ/–/ɜː/, and /ɪ/–/e/ in the /bVb/ context.
Discrimination Predictions
We compared the acoustic characteristics of English and Greek vowels to form predic-
tions about the discrimination of each L2 contrast. Eleven female adult Standard Modern
Greek speakers were recorded producing the Greek vowels embedded in a /pVs/ context (5
vowels × 4 repetitions × 11 speakers = 220 productions). The words were part of the carrier
phrase ‘Léne < target word > tóra’ (‘they say < target word > now’). In addition, 10 adult
female Standard Southern British English speakers were recorded producing the 11 Eng-
lish vowels which were part of /hVd/ words (11 vowels × 2 repetitions × 10 speakers = 220
productions). The carrier phrase was ‘they say < target word > now’. All speakers were
instructed to produce the stimuli as if speaking to a friend and were recorded at a 44.1kHz
sampling rate in quiet rooms. Only female speakers were used to eliminate the effect of
sex on the productions. Figure1 presents the F1 × F2 (measured in Hz) of Greek and Eng-
lish vowels as produced by native speakers of these languages. Figures2 and 3 present
Fig. 1 F1 × F2 (in Hz) of Greek
and English vowels as produced
by the respective native speakers
Fig. 2 Duration of Greek vowels
(in ms) as produced by Greek
native speakers
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 6 of 18
the durations (measured in ms) and standard deviations (SD) of Greek and English vowels
respectively as produced by native speakers of these languages.
According to the acoustic characteristics of English and Greek vowels, the following
predictions can be formed about the discrimination of the target English contrasts by Greek
speakers of English. These are based on the overlapping parameter threefold proposed by
UPM:
• /iː/ –/ɪ/: English /iː/ is expected to be classified as Greek /i/ as it comprises its closer
acoustic exemplar. English /ɪ/ is between Greek /i/ and /e/. However, its duration is
close to the duration of English /iː/ and, therefore, it is expected to bemainly classi-
fied as Greek /i/. Thus, a completely overlapping contrast is expected, which will signal
poor discrimination.
• /iː/–/e/: English /iː/ is expected to be classified as Greek /i/, while English /e/ will prob-
ably beclassified to its closest exemplar, that is, Greek /e/. The contrast will be nono-
verlapping, presenting excellent discrimination.
• /ɑː/–/ʌ/: English /ɑː/ lies between Greek /a/ and /o/, and thus it may overlap with both
Greek vowels. However, its duration is closer to Greek /a/ and thus it may be classified
in terms of that Greek vowel. English /ʌ/ might be classified as Greek /a/, which is its
closest acoustic exemplar. The contrast is expected to be completely overlapping.
• /æ/–/ɑː/: English /æ/ is expected to be classified as Greek /a/, while /ɑː/ might be clas-
sified as Greek /a/. Complete overlap is expected.
• /ɔː/–/ɒ/: English /ɔː/ is close to Greek /u/, but its duration is closer to Greek /o/. There-
fore, it might be an instance of Greek /o/ and /u/. English /ɒ/ is close to Greek /o/ and
it might be classified in terms of that Greek category. The contrast is expected to be
partially overlapping, exhibiting better discrimination than the completely overlapping
contrasts but worse discrimination than the nonoverlapping contrast.
• /uː/–/ʊ/: English /uː/ is acoustically close to several Greek vowels. It is expected to be
classified mostlyas Greek /o/ or /u/ since it shares more acoustic properties with these
vowels. English /ʊ/ is between Greek /o/ and /u/ but its duration is closer to the duration
of Greek /u/. The contrast might be either completely or partially overlapping.
Predictions About theEffect ofPSTM andNonverbalIQ
Although it is difficult to develop predictions for each English contrast, we hypothesize that
both PSTM and nonverbalIQ will positively affect the discrimination of all or most con-
trasts since the majority of the previous studies demonstrated either an association of pho-
netic abilities with PSTM (e.g., Aliaga-Garcia etal., 2011; Safronova, 2016) or the positive
Fig. 3 Duration of English
vowels (in ms) as produced by
English native speakers
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
Page 7 of 18 9
role of nonverbalIQ in several linguistic skills (e.g., Łockiewicz etal., 2018; Woumans,
2019) including speech perception (e.g., Georgiou, 2023a).
Hypotheses
Based on the above predictions, we aim to assess the following hypotheses:
H1 The control group (i.e., English speakers) will discriminate all but one L2 contrast
(i.e., /iː/–/e/) with higher accuracy than the experimental group (i.e., Greek speakers).
Also, evidence is expected to indicatebelowchance performance in the discrimination of
these contrasts by the experimental group, with the control group anticipated to demostrate
abovechance performance.
H2 For the Greek speakers, thediscrimination accuracy of English /iː/–/e/ will be higher
compared to the accuracy of the other contrasts, and the discrimination accuracy of /ɔː/–/ɒ/
will be higher than the accuracy of the other contrasts except /iː/–/e/.
H3 The discrimination accuracy of the L2 contrasts will be higher for Greekspeakers with
high PSTM than for Greekspeakers with low PSTM.
H4 The discrimination accuracy of the L2 contrasts will be higher for Greekspeakers with
high nonverbalIQ than for Greekspeakers with low nonverbalIQ.
Methodology
Participants
Twenty monolingually-raised Standard Modern Greek speakers with an age range of 25–45
(Mage = 32.15, SD = 7.06) (nfemales = 12) participated in the study. All participants originated
from moderate-income families and reported that they were permanent residents of Ath-
ens, Greece. None of them had ever lived for a long time in an English-speaking country.
According to their self-reports, they had knowledge of English at the B2/C1 levels and the
mean of their understanding skills in English was 4.2/5 (SD = 0.68). The mean onset age
of learning English was 7.75 years (SD = 1.73) and the daily use of English was 0.85h
(SD = 1.35) on average. The participants were divided into two groups according to their
PSTM and IQ capacities (i.e., high/low PSTM, high/low IQ) after conducting a median
split on the raw scores of the PSTM and nonverbalIQ tests. The control group consisted
of 10 speakers of Standard Southern British English with an age range of 24–42, who per-
manently resided in the UK (Mage = 31.4, SD = 6.02) (nfemales = 5). These participants had a
moderate socioeconomic status. All participants had healthy vision and hearing.
Stimuli
The stimuli consisted of the 11 English monophthongs embedded in monosyllabic /hVd/
words, which were part of the carrier phrase “They say < word > now”. These words were
hid, heed, head, herd, had, hard, hud, hod, hoard, who’d, and hood, representing the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 8 of 18
vowels /ɪ iː e ɜː æ ɑː ʌ ɒ ɔː uː ʊ/ respectively. Two adult female Standard Southern British
English native speakers were recorded at a 44.1kHz sampling rate producing the carrier
phrases. The speakers were instructed to produce the phrases as naturally as possible as if
speaking to a friend. The stimuli were normalized for peak intensity in Praat (Boersma &
Weenink, 2022).
Procedure
Discrimination Test
All participants completed an AX discrimination test in Praat. The stimuli were grouped
into six “different” pairs and six “same” pairs. The “different” pairs included the six Eng-
lish contrasts under investigation, that is, /iː/–/ɪ/, /iː/–/e/, /ɑː/–/ʌ/, /æ/–/ɑː/, /ɔː/–/ɒ/, and /
uː/–/ʊ/. Each of the six “different” conditions contained eight repetitions of the contrastive
vowels (4 AB and 4 BA types). Similarly, each of the six “same” conditions contained
eight repetitions of the contrastive vowels (4 AA and 4 BB types). A total number of 96
items (6 contrasts × 2 conditions × 8 repetitions) were discriminated by each participant and
all of them were presented in random order. The stimulus pairs always included record-
ings from different talkers. Participants were asked to sit in front of a PC, maintaining a
consistent distance from it. They listened to the stimuli through a set of headphones con-
nected to a PC and were asked to select whether the pair tokens were acoustically different
or the same by clicking on the relevant script label. The interstimulus interval was 300ms.
During the experiment, the stimuli could not be repeated and no feedback was given on
the participants’ responses. In addition, there was an optional five-minute break at the mid-
point. Prior to the main experiment, participants completed a familiarization test with four
items on the script to ensure that they understood the requirements of the test. The test
lasted about 15–20min for each participant.
PSTM Tests
Participants completed a digit span testto measure their PSTM capacities. The digit span
test is typically used to measure PSTM together with other tests such as nonword repeti-
tion tests (Brunfaut & Révész, 2015; Perez, 2020). The former has been suggested tocarry
semantic information that can bias the performance of PSTM memory (Jacquemot & Scott,
2006), while the latter is not associated with such effects as it uses pseudowords. Neverthe-
less, even when employing nonwords (or pseudowords), it is important to recognize that
these linguistic constructs are not entirely devoid of meaning and can still evoke semantic
associations (Chuang etal., 2021). For this study, the digit span test was preferred because
it has been widely used for the estimation of PSTM and it is relatively quick to administer,
making it a convenient tool to assess PSTM capacity in a short amount of time. It is also
easy to use and understand compared to nonword recognition/repetition tests. The test was
completed on a PC in quiet rooms and required the subjects to type the sequence of dig-
its they listened to from the PC loudspeakers; all of them used headphones. Participants
took a seat in front of a PC monitor, maintaining an approximate distance of 1m from it.
Apart from listening to the digits, they could also see them on the monitor. The digits were
spoken at a rate of approximately one digit per 1000ms. Each sequence began with three
digits and upon the successful record of the digits, the sequence increased in length by one
digit. There were two different digit span subtests: the forward digit span (FDS) and the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
Page 9 of 18 9
backward digit span (BDS) tests. In FDS, participants were asked to type the sequence of
digits in the order originally presented, while in BDS, they typed the sequence of digits in
the reverse order from the original presentation. For both subtests, upon an unsuccessful
record of the correct digits by the participants, another chance was given to them and they
continued from the sequence they left. Upon a second incorrect attempt, the test stopped.
The score of the participants was the sum of the number of digits correctly recorded in
FDS and BDS. Each individual took 10–15min to complete the test.
NonverbalIQ Tests
Participants’ nonverbal IQwas measured through Raven’s Standard Progressive Matrices
test (Raven etal., 2000). The test was completed in quiet rooms individually, following the
PSTM test. Participants were presented with a series of matrices or visual patterns, each
of which had one element missing. They were anticipated to identify the missing element
from the set of options provided; there were either six or eight options for each item. They
did this by discerning the underlying rules, patterns, and relationships within the matrix
and then circling the option that logically completed the pattern. A total number of 60
black and white items in five sets (e.g., A to E) of 12 items (e.g., A1–A12) were presented.
The items within a set progressively became more complex, requiring a greater amount of
cognitive capacity to encode and analyze. The test was completed within 40–45min. The
performance of the participants was measured using the raw scores.
Statistical Analysis
Bayesian regression models were used to analyse the data. The statistical analysis took
place in R (R Core Team, 2022) with the use of the brms package (Bürkner, 2017). There
are many advantages to using a Bayesian model including its ability to deal with small sam-
ples of participants (see Escudero etal., 2020; Georgiou, 2023b;van de Schoot & Depaoli,
2014). Approximate leave-one-out (LOO) cross validation was conducted to select the
best-fitting model by comparing models with different fixed and random factors. For the
examination of the discrimination accuracy, the final model (Model 1) included contrast
(six English vowel contrasts), group (experimental/control), and contrast × group as fixed
factors and subjects, and contrast and group within subjects as random factors. The final
model for PSTM and nonverbalIQ (Model 2) included contrast, group.PSTM (high/low),
group.IQ (high/low), and contrast × group as fixed factors and subjects, and contrast and
group.IQ within subjects as random factors. Weakly informative priors were used, namely,
student’s t-distribution with 3 degrees of freedom, a mean of 0, and a standard deviation
of 2.5 (see Gelman etal., 2014, 2015). We employed the Bernoulli distribution since the
dependent variable response was dichotomous (0 = incorrect, 1 = cor rect).
After fitting the models, we proceeded with hypothesis testing. The likelihood of
the test hypothesis against its alternative was estimated through the consideration of
Evidence Ratio (ER). According to the evidence categories for the Bayes Factor BF12
of Jeffreys (1961) as cited in Andraszewicz etal. (2015), anER of > 100 represents
extreme evidence, ER of 30–100 very strong evidence, ER of 10–30 strong evidence,
ER of 3–10 moderate evidence, and ER of 1–3 anecdotal evidence for a given hypothe-
sis. AnER of 1 shows no evidence at all. AnER of 1/3–1 represents anecdotal evidence,
ER of 1/10–1/3 moderate evidence, ER of 1/30–1/10 strong evidence, ER of 1/100–1/30
very strong evidence, and ER of < 1/100 extreme evidence against a hypothesis. For
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 10 of 18
practical reasons, we will consider an ER of > 10 and an ER of < 1/10 (or 0.1) as strong
evidence for and strong evidence against a particular hypothesis respectively. Apart
from ERs, we also report the posterior probabilities (PP).
Results
Discrimination
The results of the discrimination test revealed that the experimental group exhibited
poor discrimination against all but one L2 English contrast. The least discriminable
contrast was /ɪ/–/iː/, followed by /æ/–/ɑː/, /uː/–/ʊ/, /ɑː/–/ʌ/, and /ɔː/–/ɒ/. The /iː/–/e/
contrast was discriminated with high accuracy. The control group discriminated all con-
trasts with higher accuracy than the experimental group. Figure4 illustrates the accu-
racy percentages for the discrimination of the six English vowel contrasts by the control
and experimental groups.
To investigate whether there isevidence for above chance discrimination of the L2
contrasts by Greek and English speakers, we used two Bayesian regression models.
With respect to Greek speakers, there was strong evidence of above chance discrimina-
tion performance only for /iː/–/e/ and /ɔː/ – /ɒ/ (ER > 116.65, PP >0.99). With respect
to English speakers, there was strong evidence that all contrasts were discriminated
above chance (ER > 499, PP = 1). Furthermore, for the examination of H1, we fitted a
new Bayesian regression model, namely, Model 1, and conducted hypothesis testing.
As shown in Table1, there was strong evidence that all L2 contrasts were discriminated
better by the control than the experimental group (ER > 104.26, PP > 0.99).
We also compared the discrimination accuracy of different L2 contrasts within the
experimental group (H2) using further hypothesis testing. The results demonstrated
strong evidence that English /iː/–/e/ was discriminated with higher accuracy than four
contrasts (ER = inf, PP = 1) and that /ɪ/–/iː/ was discriminated with lower accuracy than
/iː/–/e/ (ER = 0, PP = 0). Also, there was strong evidence that English /ɔː/–/ɒ/ was dis-
criminated with higher accuracy than /uː/–/ʊ/ (ER = inf, PP = 1) and that /ɪ/–/iː/ and
/æ/–/ɑː/ were discriminated worse than /ɔː/–/ɒ/ (ER < 0.04, PP < 0.04). In sum, the
results show that /iː/–/e/ was discriminated more accurately than all the other contrasts,
while /ɔː/–/ɒ/ was discriminated better than three contrasts and worse than onecontrast,
namely /iː/–/e/. The results of hypothesis testing for the comparison of the accuracy of
different L2 contrasts as discriminated by the Greek speakers are shown in Table2.
Fig. 4 Discrimination accuracy
of L2 English contrasts (percent-
age of correct responses) by the
control (left) and the experimen-
tal (right) groups. The error bars
show the SEs and the dashed line
shows chance performance
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
Page 11 of 18 9
PSTM
For purposes of testing whether the high PSTM group discriminated the L2 contrasts
more accurately than the low PSTM group (H3), we fitted another Bayesian regression
model, namely, Model 2. The results of hypothesis testing showed that there was no evi-
dence for better discrimination of five L2 contrasts by the highPSTM group (ER < 5.07,
PP < 0.84). The results of hypothesis testing are shown in Table3.
Table 1 Hypothesis testing results for ERs of the contrast × group interaction in Model 1
The hypothesis tests whether the control group discriminated each L2 contrast more accurately than the
experimental group
Hypothesis con-
trol> experimental
Estimate Estimate Error CI. Lower CI. Upper ER PP
/ɪ/–/iː/ 3.57 0.63 2.57 4.63 Inf 1.00
/iː/–/e/ 3.10 1.66 0.79 6.07 104.26 0.99
/ɑː/–/ʌ/ 2.46 0.56 1.57 3.43 3999.00 1.00
/æ/–/ɑː/ 3.02 0.55 2.14 3.95 Inf 1.00
/ɔː/–/ɒ/ 2.12 0.59 1.18 3.10 3999.00 1.00
/uː/–/ʊ/ 2.84 0.60 1.88 3.88 Inf 1.00
Table 2 Hypothesis testing results for the ERs of the contrast effect in Model 1
The hypothesis tests whether each L2 contrast was discriminated more accurately than any other by the
Greek speakers
Hypothesis
(contrasts)
Estimate Estimate Error CI. Lower CI. Upper ER PP
1 > 2 −3.67 0.50 −4.51 −2.90 0.00 0.00
1 > 3 −0.50 0.33 −1.04 0.03 0.07 0.06
1 > 4 −0.31 0.33 −0.86 0.24 0.21 0.17
1 > 5 −0.89 0.34 −1.44 −0.36 0.00 0.00
1 > 6 −0.32 0.33 −0.87 0.24 0.20 0.16
2 > 3 3.17 0.50 2.39 4.03 Inf 1.00
2 > 4 3.37 0.50 2.59 4.23 Inf 1.00
2 > 5 2.78 0.50 2.00 3.63 Inf 1.00
2 > 6 3.36 0.50 2.58 4.22 Inf 1.00
3 > 4 0.20 0.34 −0.35 0.76 2.67 0.73
3 > 5 −0.39 0.34 −0.95 0.18 0.14 0.12
3 > 6 0.18 0.33 −0.35 0.73 2.51 0.71
4 > 5 −0.59 0.34 −1.13 −0.03 0.04 0.04
4 > 6 −0.01 0.33 −0.55 0.53 0.93 0.48
5 > 6 3.43 0.70 2.31 4.61 Inf 1.00
---
1 = /ɪ/–/iː/, 2 = /iː/–/e/, 3 = /ɑː/–/ʌ/, 4 = /æ/–/ɑː/, 5 = /ɔː/–/ɒ/, 6 = /uː/–/ʊ/
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 12 of 18
NonverbalIQ
We conducted another hypothesis testing to estimate whether the high IQ group discrimi-
nated the L2 contrasts more accurately than the low IQ group (H4). We found strong
evidence that /ɪ/–/iː/, /ɑː/–/ʌ/, /æ/–/ɑː/, and /ɔː/–/ɒ/ (ER > 11.86, PP > 0.92) were dis-
criminated more accurately by the high IQ group compared to the low IQ group. Weaker
evidence was found for two L2 contrasts (ER = 3.04–3.96, PP = 0.75–0.80). The results of
hypothesis testing are shown in Table4.
Discussion
The present study investigated the discrimination of L2 English vowel contrasts by Greek
speakers. It also examined the role of PSTM and nonverbalIQ in the discrimination of
L2 contrasts. The participants were adult Greek speakers with experience in English.
They completed three tests: an AX discrimination, a digit span, and an intelligence test in
controlled environments. The results of all tests were analyzed with the use of Bayesian
regression models. Apart from revealing the speech perception patterns of speakers with an
underresearched L1, that is Greek, this study provided further evidence—in the relatively
scarce literature—about the effect of PSTM and nonverbalIQ on L2 speech perception.
Table 3 Hypothesis testing results for the contrast × group.PSTM interaction in Model 2
The hypothesis tests whether the high PSTM group discriminated each L2 contrast more accurately than the
low PSTM group
Hypothesis
high > low
Estimate Estimate Error CI.Lower CI.Upper ER PP
/ɪ/–/iː/ −0.35 0.57 −1.30 0.59 0.37 0.27
/iː/–/e/ −0.36 0.91 −1.86 1.11 0.52 0.34
/ɑː/–/ʌ/ −0.80 0.59 −1.76 0.13 0.09 0.08
/æ/–/ɑː/ −0.44 0.53 −1.30 0.42 0.27 0.21
/ɔː/–/ɒ/ −0.41 0.59 −1.40 0.53 0.32 0.24
/uː/–/ʊ/ 0.56 0.58 −0.40 1.55 5.07 0.84
Table 4 Hypothesis testing results for the contrast × group.IQ interaction in Model 2
The hypothesis tests whether the high IQ group discriminated each L2 contrast more accurately than the
low IQ group
Hypothesis
high > low
Estimate Estimate Error CI.Lower CI.Upper ER PP
/ɪ/–/iː/ 0.90 0.58 −0.06 1.85 14.81 0.94
/iː/–/e/ 0.78 0.91 −0.68 2.33 3.96 0.80
/ɑː/–/ʌ/ 0.81 0.57 −0.11 1.74 11.86 0.92
/æ/–/ɑː/ 1.00 0.50 0.18 1.83 41.55 0.98
/ɔː/–/ɒ/ 1.34 0.59 0.40 2.33 110.11 0.99
/uː/–/ʊ/ 0.37 0.55 −0.56 1.26 3.04 0.75
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
Page 13 of 18 9
Greek speakers of L2 English have insufficient abilities to discriminate particular Eng-
lish vowel contrasts. This finding confirms H1 since we found strong evidence that the
control group discriminated all L2 contrasts better than the experimental group. Also,
we found strong evidence that all but two contrasts were discriminated by Greek speak-
ers below chance, while English speakers discriminated all contrasts above chance. The
confirmation of this hypothesis underlines the great effect of speakers’ L1 on L2 vowel
discrimination since every L2 sound is filtered through the speakers’ native inventory. One
explanation would be that the Greek vowel system is smaller and less complex than the
English vowel system; therefore, two or more L2 sounds might have been accommodated
to a single L1 category, creating perceptual challenges regarding the discrimination of
these contrasts (for further evidence, see Fox etal., 1995; Iverson & Evans, 2007, 2009).
An additional explanation would posit that the acousticophonetic differences between the
vowels of the two languages interfered with the perception of the L2 sounds, leading to
perceptual problems. The poor performance of the speakers can also beinterpreted as a
result ofthe infrequent use of English in their daily life and the limited exposureto L2 nat-
uralistic stimuli, which is considered an important factor in improving learners’ L2 speech
perception abilities (Georgiou, 2021b). Usually, active learners of an L2 (e.g., see Geor-
giou etal., 2020b) or individuals with naturalistic access to the L2 speech (e.g., see Geor-
giou, 2018) are characterized by better discrimination abilities due to more robust mental
representations compared to nonactive learners and speakers who live in a context where
the L2 is not dominant.
H2 was also confirmed. This hypothesis was based on previous evidence which sug-
gested that the consideration of the acoustic properties of L1 and L2 vowels can predict to
some extent the speakers’ L2 discrimination patterns. Indeed, our initial predictions devel-
oped on the basis of crosslinguistic acoustic similarity were verified by the discrimina-
tion results. Specifically, English /iː/–/e/, which was regarded as a nonoverlapping contrast,
had higher discrimination accuracy than the other L2 contrasts, and English /ɔː/–/ɒ/, which
was a partially overlapping contrast, had higher discrimination accuracy than the other L2
contrasts except /iː/–/e/. Therefore, crosslinguistic acoustic similarity can somehow besuc-
cessful in predicting the discrimination accuracy of the L2 contrasts (see Alispahic etal.,
2017; Elvin etal., 2014; Escudero etal., 2012; Georgiou, 2024). In addition, the overlap-
ping parameters proposed by UPM can be a good metric for the estimation of L2 contrast
discrimination accuracy (see Georgiou, 2022b).
Against our expectations, there was no evidence suggesting that speakers with high PSTM
discriminate particular L2 English vowel contrasts more accurately compared to speakers with
low PSTM. Therefore, H3 was rejected. This is inconsistent with several studies that have found
positive effects of PSTM on L2 speech perception (e.g., Lengeris & Nicolaides, 2014; MacKay
etal., 2001). However, it agrees with the findings of other studiesthat providecounterevidence.
Safronova and Mora (2012) observed that Spanish/Catalan learners of English with a high
PSTM capacity could not perceive the English contrast /ɪ/–/iː/ more accurately than learners
with a low PSTM capacity. Also, Ghaffarvand Mokari and Werner (2019) reported that PSTM
was not associated with gains of phonetic training in the discrimination of English vowel pairs.
Our findings may have several explanations. For example, the outcomes may depend on the
type of test used to assess speakers’ PSTM. In our study, we used a digit span test, while other
studies reporting a positive effect of PSTM used nonword recognition or nonword repetition
tasks. This result may have emerged from the fact that nonword tasks are more specific for the
measurement of phonological processing and phonological information stored in PSTM com-
pared to the digit span tests, which may additionallydemand WM skills (especially in BDS)
(Volpato, 2020). Notably, WM capacities were not found to be associated with L2 speech
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 14 of 18
perception in some recent studies (e.g., Inceoglu, 2019), providing a justification for the nonex-
istent effect of PSTM on L2 speech perception in this study, which employed a digit span test.
In addition, the results may suggest a complex interplay between PSTM and L2 sound percep-
tion. While PSTM is recognized as a contributing factor to L2 sound perception, the scope and
character of its impact can vary significantly depending on factors such as the speakers’ L1 and
L2, and linguistic materials such as contrasts under investigation in a given study. Therefore, the
role of PSTM might not be universally consistent and may depend on the aforementioned fac-
tors. Another possible explanation for the results of this study is the role of perceived difficulty
for each contrast. Considering that the processing of difficult contrasts requires larger cognitive
demands (Ghaffarvand Mokari & Werner, 2019), a specific contrast which is perceived as dif-
ficult by anindividual with high PSTMmay be discriminated in the same manner by anindi-
vidual with low PSTMwho has perceived it as an easy contrast. Finally, other extralinguistic
and extracognitive factors such as speakers’ daily fatigue or psychological stress may have also
affected the results. This is because fatigue is associated with decreased motivation to apply
effort to the task (Wang etal., 2018). Speakers’ psychological stress during the task (as the task
was inspected live by the researchers) may have affected the participants’ performance as there
is evidence that the presence of stress affects memory functions (Schoofs etal., 2009).
The findings demonstrated that there was overall evidence that speakers with high nonver-
bal IQ discriminated the L2 English contrasts more accurately than speakers with low non-
verbal IQ. Strong evidence was found for four out of six L2 contrasts. Therefore, H4 can be
accepted to a large extent, highlighting the positive role of nonverbal IQ in L2 speech per-
ception. While these results diverge from earlier findings that showed no association between
phonetic abilities and nonverbal IQ (e.g., Rota & Reiterer, 2009), they align with the results
of Georgiou (2023a), providing further evidence supporting the connection between nonver-
bal IQ and phonetic (perceptual) abilities. NonverbalIQ relates to several cognitive functions,
which include perception, learning, and language abilities (Kiely, 2014). All these functions
apply to speech perception, which requires listeners to extract acoustic information from the
speech signal and organize speech sounds categorically in the mind. So, the advantage of
high nonverbalIQ individuals may be attributed to their enhanced ability to learn and process
information (see Stenberg, 1985), including phonological and phonetic relationships. Another
explanation is that speakers with high nonverbalIQ have increased abilities in controlling
their attention (Sweller, 1988). This means that they can potentially attune to the L2 sound
patterns more easily compared to speakers with low nonverbalIQ. For all these reasons, the
link between nonverbal IQ and speech perception seems reasonable and speakers with high
nonverbalIQ are likely to perform better in speech perception tasks than speakers with low
nonverbalIQ. We propose that the effect of nonverbalIQ on L2 phonological discrimination
is contrast-specific since not all contrasts were found to be affected to the same extent. At this
stage, we are not able to explain why nonverbalIQ did not affect the discrimination of particu-
lar contrasts. Perhaps, this is related to the level of the perceived difficulty of each contrast,
just like in the case of PSTM, which may prevent speakers with high nonverbalIQ from per-
ceiving acoustic differences better than speakers with low nonverbalIQ.
Conclusions
Our findings suggest that Greek speakers demonstrate poor performance in the discrimi-
nation of the majority of L2 Englishcontrasts. In addition, PSTM did not have any effect
on L2 speech perception, while there was an effect of nonverbal IQ on most contrasts.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
Page 15 of 18 9
Although we attempted to control several factors such as L2 proficiency, L2 use, age of L2
learning onset etc. other factors such as speakers’ motivation, attention control, phonetic
aptitude, etc. were not considered for practical reasons; these unexamined factors may have
affected the results, underscoring the need for their comprehensive examination in subse-
quent studies. Also, we used a single tool to measure PSTM and nonverbalIQ capacities.
While these measures provided valuable insights into the association between cognitive
functions and L2 speech perception, incorporating multiple tools and assessments would
offer a more comprehensive understanding of this relationship. For example, a future study
can include additional measures of PSTM and nonverbalIQ using a variety of tools such as
nonword repetition and nonword recognition tasks and the WASI test respectively. Finally,
considering the varying effects of cognitive measures on each L2 contrast, future studies
may explain why particular L2 contrasts are affected differently by measures of cognitive
functions.
Acknowledgements This work was supported by the Phonetic Lab and the Cyprus Linguistics and Humani-
ties Research Group of the University of Nicosia.
Funding Open access funding provided by the Cyprus Libraries Consortium (CLC). No fundings sources
are available.
Data Availability Not applicable.
Declarations
Conflict of interest We have no conflicts of interest to disclose.
Ethical Approval This study received ethical approval from the Department of Languages and Literature of
the University of Nicosia. All participants were informed about the goals of this study and gave their written
consent for participation, according to the Declaration of Helsinki.
Informed Consent Participants gave their written consent for their participation in the experiments.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Com-
mons licence, and indicate if changes were made. The images or other third party material in this article
are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly
from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
References
Aliaga-García, C., Mora, J. C., & Cerviño-Povedano, E. (2011). L2 speech learning in adulthood and pho-
nological short-term memory. Poznań Studies in Contemporary Linguistics, 47(1), 1–14.
Alispahic, S., Mulak, K. E., & Escudero, P. (2017). Acoustic properties predict perception of unfamiliar Dutch
vowels by adult Australian English and Peruvian Spanish listeners. Frontiers in Psychology, 8, 52.
Andraszewicz, S., Scheibehenne, B., Rieskamp, J., Grasman, R., Verhagen, J., & Wagenmakers, E. J.
(2015). An introduction to Bayesian hypothesis testing for management research. Journal of Manage-
ment, 41(2), 521–543.
Baddeley, A., Lewis, V., & Vallar, G. (1984). Exploring the articulatory loop. The Quarterly Journal of
Experimental Psychology Section A, 36(2), 233–252.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 16 of 18
Best, C. T., & Tyler, M. D. (2007). Nonnative and second-language speech perception: Commonalities and
complementarities. In O.-S. Bohn & M. J. Munro (Eds.), Language experience in second language
speech learning: In honor of James Emil Flege (pp. 13–34). John Benjamins.
Boersma, P., & Weenink, D. (2022). Praat: doing phonetics by computer [Computer program]. Retrieved
from http:// www. fon. hum. uva. nl/ praat/
Brunfaut, T., & Révész, A. (2015). The role of task and listener characteristics in second language listen-
ing. TESOL Quarterly, 49, 141–168.
Bürkner, P. C. (2017). brms: An R package for Bayesian multilevel models using Stan. Journal of Statis-
tical Software, 80, 1–28.
Carroll, J. (1962). The prediction of success in intensive foreign language training. In R. Glaser (Ed.),
Training and research in education (pp. 87–136). Prentice-Hall.
Cattell, R. B. (1963). Theory of fluid and crystallized intelligence: A critical experiment. Journal of
Educational Psychology, 54, 1–22.
Cebrian, J. (2006). Experience and the use of non-native duration in L2 vowel categorization. Journal of
Phonetics, 34(3), 372–387.
Cerviño-Povedano, E. and Mora, J. C. (2010). Investigating Catalan learners of English over-reliance on
duration: Vowel cue weighting and phonological short-term memory. Proceedings of New Sounds
2010, pp. 53–64.
Christiner, M., Rüdegger, S., & Reiterer, S. M. (2018). Sing Chinese and tap Tagalog? Predicting indi-
vidual differences in musical and phonetic aptitude using language families differing by sound-
typology. International Journal of Multilingualism, 15(4), 455–471.
Chuang, Y. Y., Vollmer, M. L., Shafaei-Bajestan, E., Gahl, S., Hendrix, P., & Baayen, R. H. (2021). The
processing of pseudoword form and meaning in production and comprehension: A computational
modeling approach using linear discriminative learning. Behavior Research Methods, 53(3), 945–976.
Daller, M., & Ongun, Z. (2018). The threshold hypothesis revisited: Bilingual lexical knowledge and
non-verbal IQ development. International Journal of Bilingualism, 22(6), 675–694.
Dryden, A., Allen, H. A., Henshaw, H., & Heinrich, A. (2017). The association between cognitive per-
formance and speech-in-noise perception for adult listeners: A systematic literature review and
meta-analysis. Trends in Hearing, 21, 1–21.
Elvin, J., Escudero, P., & Vasiliev, P. (2014). Spanish is better than English for discriminating Portu-
guese vowels: Acoustic similarity versus vowel inventory size. Frontiers in Psychology, 5, 1188.
Escudero, P., Ellen, S., & Mitterer, H. (2012). The perception of English front vowels by North Holland
and Flemish listeners: Acoustic similarity predicts and explains cross-linguistic and L2 perception.
Journal of Phonetics, 40(2), 280–288.
Escudero, P., Jones Diaz, C., Hajek, J., Wigglesworth, G., & Smit, E. A. (2020). Probability of heritage
language use at a supportive early childhood setting in Australia. Frontiers in Education, 5, 93.
Evans, B. G., & Alshangiti, W. (2018). The perception and production of British English vowels and
consonants by Arabic learners of English. Journal of Phonetics, 68, 15–31.
Flege, J. E., & Fletcher, K. L. (1992). Talker and listener effects on degree of perceived foreign accent.
The Journal of the Acoustical Society of America, 91(1), 370–389.
Flege, J. E., & Liu, S. (2001). The effect of experience on adults’ acquisition of a second language. Stud-
ies in Second Language Acquisition, 23(4), 527–552.
Flege, J. E., Frieda, E. M., & Nozawa, T. (1997). Amount of native-language (L1) use affects the pro-
nunciation of an L2. Journal of Phonetics, 25(2), 169–186.
Flege, J. E., Munro, M. J., & MacKay, I. R. (1995). Factors affecting strength of perceived foreign accent
in a second language. The Journal of the Acoustical Society of America, 97(5), 3125–3134.
Fox, R. A., Flege, J. E., & Munro, M. J. (1995). The perception of English and Spanish vowels by native
English and Spanish listeners: A multidimensional scaling analysis. The Journal of the Acoustical
Society of America, 97(4), 2540–2551.
French, L., & O’Brien, I. (2008). Phonological memory and children’s second language grammar learn-
ing. Applied Psycholinguistics, 29, 463–487.
Gelman, A., Hwang, J., & Vehtari, A. (2014). Understanding predictive information criteria for Bayesian
models. Statistics and Computing, 24(6), 997–1016.
Gelman, A., Lee, D., & Guo, J. (2015). Stan: A probabilistic programming language for Bayesian infer-
ence and optimization. Journal of Educational and Behavioral Statistics, 40(5), 530–543.
Georgiou, G. P. (2018). Discrimination of L2 Greek vowel contrasts: Evidence from learners with Ara-
bic L1 background. Speech Communication, 102, 68–77.
Georgiou, G. P. (2021a). Toward a new model for speech perception: The Universal Perceptual Model
(UPM) of Second Language. Cognitive Processing, 22(2), 277–289.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
Page 17 of 18 9
Georgiou, G. P. (2021b). Effects of phonetic training on the discrimination of second language sounds
by learners with naturalistic access to the second language. Journal of Psycholinguistic Research,
50(3), 707–721.
Georgiou, G. P. (2022a). The impact of auditory perceptual training on the perception and production of English
vowels by Cypriot Greek children and adults. Language Learning and Development, 18(4), 379–392.
Georgiou, G. P. (2022b). The acquisition of/ɪ/–/iː/ Is challenging: Perceptual and production evidence from
Cypriot Greek speakers of english. Behavioral Sciences, 12(12), 469.
Georgiou, G. P. (2023a). Nonverbal intelligence does matter for the perception of second language sounds.
Psych, 5(1), 123–132.
Georgiou, G. P. (2023b). Bayesian models are better than frequentist models in identifying differences in small
datasets comprising phonetic data. arXiv (Cornell University). https:// doi. org/ 10. 48550/ arxiv. 2312. 01146
Georgiou, G. P. (2024). Classification of English vowels in terms of Cypriot Greek categories: The role of
acoustic similarity between L1 and L2 sounds. Canadian Journal of Linguistics. https:// doi. org/ 10. 1017/
cnj. 2024.1
Georgiou, G. P. (2023c). Speakers of different L1 dialects with acoustically proximal vowel systems present
with similar nonnative speech perception abilities: Data from Greek listeners of Dutch. Speech Communi-
cation, 150, 32–40.
Georgiou, G. P. (2023d). Comparison of the prediction accuracy of machine learning algorithms in crosslinguis-
tic vowel classification. Scientific Reports, 13, 15594.
Georgiou, G. P., & Dimitriou, D. (2023). Perception of Dutch vowels by Cypriot Greek listeners: To what extent
can listeners’ patterns be predicted by acoustic and perceptual similarity? Attention, Perception, and Psy-
chophysics, 85, 2459–2474.
Georgiou, G. P., Perfilieva, N., Denisenko, V., & Novospasskaya, N. (2020a). Perceptual realization of Greek
consonants by Russian monolingual speakers. Speech Communication, 125, 7–14.
Georgiou, G. P., Perfilieva, N., & Tenizi, M. (2020b). Vocabulary size leads to a better attunement to L2 phonetic
differences: Clues from Russian learners of English. Language Learning and Development, 16(4), 382–398.
Georgiou, G. P., & Themistocleous, C. (2021). Vowel learning in diglossic settings: Evidence from Arabic-
Greek learners. International Journal of Bilingualism, 25(1), 135–150.
Ghaffarvand Mokari, P., & Werner, S. (2019). On the role of cognitive abilities in second language vowel learn-
ing. Language and Speech, 62(2), 260–280.
Inceoglu, S. (2019). Individual differences in L2 speech perception: The role of phonological memory and
lipreading ability. The Modern Language Journal, 103(4), 782–799.
Iverson, P., & Evans, B. G. (2007). Learning English vowels with different first-language vowel systems: Per-
ception of formant targets, formant movement, and duration. The Journal of the Acoustical Society of
America, 122(5), 2842–2854.
Iverson, P., & Evans, B. G. (2009). Learning English vowels with different first-language vowel systems II:
Auditory training for native Spanish and German speakers. The Journal of the Acoustical Society of Amer-
ica, 126(2), 866–877.
Iverson, P., Kuhl, P. K., Akahane-Yamada, R., Diesch, E., Kettermann, A., & Siebert, C. (2003). A perceptual
interference account of acquisition difficulties for non-native phonemes. Cognition, 87(1), B47–B57.
Jacquemot, C., & Scott, S. K. (2006). What is the relationship between phonological short-term memory and
speech processing? Trends in Cognitive Sciences, 10(11), 480–486.
Jeffreys, H. (1961). The Theory of Probability. Oxford University Press.
Kiely, K. M. (2014). Cognitive function. Encyclopedia of Quality of Life and Well-Being Research, 508,
974–978.
Kim, Y., Tracy-Ventura, N., & Jung, Y. (2016). A measure of proficiency or short-term memory? Validation of
an elicited imitation test for SLA research. The Modern Language Journal, 100(3), 655–673.
Lengeris, A. (2009). Perceptual assimilation and L2 learning: Evidence from the perception of Southern British
English vowels by native speakers of Greek and Japanese. Phonetica, 66, 169–187.
Lengeris, A., & Nicolaidis, K. (2014). English consonant confusions by Greek listeners in quiet and noise and
the role of phonological short-term memory. In Proceedings of the 15th Annual Conference of the Interna-
tional Speech Communication Association (Interspeech), Singapore, pp. 534–538
Łockiewicz, M., Sarzała-Przybylska, Z., & Lipowska, M. (2018). Early predictors of learning a foreign language
in pre-school–Polish as a first language, English as a foreign language. Frontiers in Psychology, 9, 1813.
MacKay, I. R. A., Meador, D., & Flege, J. E. (2001). The identification of English consonants by native speak-
ers of Italian. Phonetica, 58, 103–125.
Martin, K. I., & Ellis, N. C. (2012). The roles of phonological STM and working memory in L2 grammar and
vocabulary learning. Studies in Second Language Acquisition, 34(3), 379–413.
Masoura, E. V., & Gathercole, S. E. (2005). Phonological short-term memory skills and new word learning in
young Greek children. Memory, 13, 422–429.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Journal of Psycholinguistic Research (2024) 53:9
1 3
9 Page 18 of 18
Mayr, R., & Escudero, P. (2010). Explaining individual variation in L2 perception: Rounded vowels in English
learners of German. Bilingualism: Language and Cognition, 13(3), 279–297.
Meador, D., Flege, J. E., & MacKay, I. R. (2000). Factors affecting the recognition of words in a second lan-
guage. Bilingualism: Language and Cognition, 3(1), 55–67.
Montgomery, J. W., Polunenko, A., & Marinellie, S. A. (2009). Role of working memory in children’s under-
standing spoken narrative: A preliminary investigation. Applied psycholinguistics, 30(3), 485–509.
O’Brien, I., Segalowitz, N., Collentine, J., & Freed, B. (2006). Phonological memory and lexical, narrative, and
grammatical skills in second language oral production by adult learners. Applied Psycholinguistics, 27,
377–402.
Perez, M. M. (2020). Incidental vocabulary learning through viewing video: The role of vocabulary knowledge
and working memory. Studies in Second Language Acquisition, 42(4), 749–773.
R Core Team (2022). R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. https:// www.R- proje ct. org/
Raven, J., Raven, J. C., & Court, H. H. (2000). Raven manual: Section3. Oxford Psychologists Press Ltd.
Rota, G., & Reiterer, S. (2009). Cognitive aspects of pronunciation talent. Language Talent and Brain Activity, 1,
67–96.
Safronova, E. (2016). The role of cognitive ability in the acquisition of second language perceptual phonologi-
cal competence (PhD thesis, University of Barcelona).
Safronova, E., & Mora, J. C. (2012). Acoustic and phonological memory in L2 vowel perception.At a time of
crisis: English and American studies in Spain, pp. 384–390.
Schmidt, L. B. (2018). L2 development of perceptual categorization of dialectal sounds: A study in Spanish.
Studies in Second Language Acquisition, 40(4), 857–882.
Schoofs, D., Wolf, O. T., & Smeets, T. (2009). Cold pressor stress impairs performance on working memory
tasks requiring executive functions in healthy young men. Behavioral Neuroscience, 123(5), 1066.
Shinohara, Y., & Iverson, P. (2018). High variability identification and discrimination training for Japanese
speakers learning English/r/–/l. Journal of Phonetics, 66, 242–251.
Sparks, R. L., Patton, J., & Ganschow, L. (2012). Profiles of more and less successful L2 learners: A cluster
analysis study. Learning and Individual Differences, 22(4), 463–472.
Sparks, R., Ganschow, L., Javorsky, J., Pohlman, J., & Patton, J. (1992). Test comparisons among hi-risk, low-
risk, and learning disabled students enrolled in foreign language courses. Modern Language Journal, 76,
142–159.
Spearman, C. (1904). The proof and measurement of association between two things. The American Journal of
Psychology, 15(1), 72–101.
Sternberg, R. J. (1985). Implicit theories of intelligence, creativity, and wisdom. Journal of Personality and
Social Psychology, 49(3), 607–627.
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2),
257–285.
Tree, J. J., & Playfoot, D. (2019). How to get by with half a loop–An investigation of visual and auditory codes
in a case of impaired phonological short-term memory (pSTM). Cortex, 112, 23–36.
Van De Schoot, R., & Depaoli, S. (2014). Bayesian analyses: Where to start and what to report. The European
Health Psychologist, 16(2), 75–84.
Volpato, F. (2020). Verbal working memory resources and comprehension of relative clauses in children with
cochlear implants. First Language, 40(4), 390–410.
Wang, Y., Naylor, G., Kramer, S. E., Zekveld, A. A., Wendt, D., Ohlenforst, B., & Lunner, T. (2018). Relations
between self-reported daily-life fatigue, hearing status, and pupil dilation during a speech perception in
noise task. Ear and Hearing, 39(3), 573.
Wechsler, D. (1999). Wechsler abbreviated scale of intelligence (WASI). Pearson Assessment.
Wechsler, D. (2008). Wechsler Adult Intelligence Scale (4th Ed.) Pearson.
Woumans, E., Ameloot, S., Keuleers, E., & Van Assche, E. (2019). The relationship between second language
acquisition and nonverbal cognitive abilities. Journal of Experimental Psychology: General, 148(7), 1169.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center
GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers
and authorised users (“Users”), for small-scale personal, non-commercial use provided that all
copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of
use (“Terms”). For these purposes, Springer Nature considers academic use (by researchers and
students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and
conditions, a relevant site licence or a personal subscription. These Terms will prevail over any
conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to
the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of
the Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may
also use these personal data internally within ResearchGate and Springer Nature and as agreed share
it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not otherwise
disclose your personal data outside the ResearchGate or the Springer Nature group of companies
unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial
use, it is important to note that Users may not:
use such content for the purpose of providing other users with access on a regular or large scale
basis or as a means to circumvent access control;
use such content where to do so would be considered a criminal or statutory offence in any
jurisdiction, or gives rise to civil liability, or is otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association
unless explicitly agreed to by Springer Nature in writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a
systematic database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a
product or service that creates revenue, royalties, rent or income from our content or its inclusion as
part of a paid for service or for other commercial gain. Springer Nature journal content cannot be
used for inter-library loans and librarians may not upload Springer Nature journal content on a large
scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not
obligated to publish any information or content on this website and may remove it or features or
functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke
this licence to you at any time and remove access to any copies of the Springer Nature journal content
which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or
guarantees to Users, either express or implied with respect to the Springer nature journal content and
all parties disclaim and waive any implied warranties or warranties imposed by law, including
merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published
by Springer Nature that may be licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a
regular basis or in any other manner not expressly permitted by these Terms, please contact Springer
Nature at
onlineservice@springernature.com