ChapterPDF Available

Second language speech learning: Theory, findings and problems

Authors:
James Emil Flege at University of Alabama at Birmingham
James Emil Flege
Download full-text PDFDownload full-text PDF
Read full-text
Download citation

Abstract and Figures

The aim of our research is to understand how speech learning changes over the life span and to explain why "earlier is better" as far as learning to pronounce a second language (L2) is concerned. An assumption we make is that the phonetic systems used in the production and perception of vowels and consonants remain adaptiive over the life span, and that phonetic systems reorganize in response to sounds encountered in an L2 through the addition of new phonetic categories, or through the modification of old ones. The chapter is organized in the following way. Several general hypotheses concerning the cause of foreign accent in L2 speech production are summarized in the introductory section. In the next section, a model of L2 speech learning that aims to account for age-related changes in L2 pronunciation is presented. The next three sections present summaries of empirical research dealing with the production and perception of L2 vowels, word-initial consonants, and word-final consonants. The final section discusses questions of general theoretical interest, with special attention to a featural (as opposed to a segmental) level of analysis. Although nonsegmental (i.e., prosodic) dimensions are an important source of foreign accent, the present chapter focuses on phoneme-sized units of speech. Although many different languages are learned as an L2, the focus is on the acquisition of English.
Figures - uploaded by James Emil Flege
Author content
All figure content in this area was uploaded by James Emil Flege
Content may be subject to copyright.
Flege_in_Strange_19
95.pdf
Content uploaded by James Emil Flege
Author content
All content in this area was uploaded by James Emil Flege on Jun 19, 2019
Content may be subject to copyright.
Flege_in_Strange_19
95.pdf
Content uploaded by James Emil Flege
Author content
All content in this area was uploaded by James Emil Flege on Jun 17, 2019
Content may be subject to copyright.
A preview of the PDF is not available
... Non-immersion beginner learners are likely to process similar-sounding vowels in English by accessing their L1 vowel categories (Flege, 1995). This results in L1 transfer in their production, but it may also lead to what is known as phonetic drift -a short-term shift in the realisation of L1 categories in response to recent L2 exposure (Chang, 2019b). ...
... We also did not find evidence for input-driven change in /i, ɪ/. In light of the SLM (Flege, 1995), the qualitative difference between them in class input may not have been substantial enough to drive the learners to notice it. ...
Preschoolers’ Production of L2 Vowels Is Affected by Input Quality: A Longitudinal Study
Article
Full-text available
  • Apr 2025
We explore the impact of in-class input on pre-literate foreign language learners’ production of second-language (L2) English vowels, comparing vowels in words newly learnt in class from a near-native speaker of Standard Southern British English (SSBE), and earlier-learnt words heard from foreign-accented speakers. We further explore if and how classroom exposure to SSBE affects learners’ first-language (L1) vowels. Over 16 weeks, we recorded seven Moravian Czech preschoolers who attended weekly English lessons. They produced words with SSBE /i, ɪ, ɛ, æ, ʌ/ and Moravian Czech /iː, i, e, a/. We modelled normalised vowel height and retraction using linear mixed effects models. In newly learnt words, participants produced L2 /ɛ/ as higher than /æ/, showing greater separation of the vowels than in earlier-learnt words. Still, /ɛ, æ/ overlap was observed, with /æ/ showing great variability. Participants raised L2 /i/ in later recordings, producing /i/ and /ɪ/ similarly in earlier-learnt and newly learnt L2 words. L1 production was recorded twice approximately nine weeks apart. Participants raised L1 /iː/ and /i/, which became more similar in height and overlapped more with L2 /i/. L1 /e/ retracted and overlapped more with L2 /æ/; L1 /a/ retracted, assimilating to L2 /ʌ/. Our findings illustrate the influence of input characteristics on child learners’ production of L2 vowels and suggest that classroom input can drive phonetic drift in young children.
View
... So far, there have been many studies on the impact of L1 on SLA, with many focusing on the transfer effects of dialect phonetics, especially negative ones [1][2][3]. As for Chinese dialects, Guan selected Shenyang local English learners as the research subjects to explore the acquisition of nine English fricatives by using speech acoustics software to extract the acoustic parameters, the center of gravity and dispersion. ...
Comparative Study on Coda/m/ Transfer Effect Between With-coda/m/ Dialect Speakers and Without-coda/m/ Dialect Speakers
Article
Full-text available
  • May 2025
There are both positive transfer and negative transfer in second language acquisition (SLA). The purpose of this paper is to explore the ways in which sharing features of L1 and L2 can contribute to L2 learning, while different features can negatively affect L2 learning. The listening test on coda/m/ was carried out on the Chinese language subjects of with-coda/m/ dialects and without-coda/m/ dialects. The results showed that: With-coda/m/ dialect speakers have no significant advantage over general coda/m/ perception over Without-coda/m/ dialect speakers. However, with-coda/m/ dialect speakers have significant advantage over the other group on certain vowel-coda/m/ sets that exist in their L1. This work provides some inspiration for English teaching, and also deepens the relevant theories of second language acquisition. In addition, for learners with specific speech disorders, this study helps to identify the source of the problem, provide data support for the field of speech pathology, and guide the design of more accurate intervention and treatment programs for speech disorders.
View
... This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). influence second-language learners' (in)ability to perceive and learn non-native sounds in target L2 (e.g., Best, 1995;Flege, 1995). Studies of non-native speech learning suggest that acoustic cues used for L2 sound contrasts are often weighted as a function of their informativeness for signaling L1 sound contrasts. ...
The role of prior knowledge in second-language learners' overnight consolidation of Cantonese tones
Article
Full-text available
  • May 2025
  • J PHONETICS
This study examines the role of prior (tonal) knowledge in memory consolidation of non-native tones after an overnight sleep. While memory consolidation is beneficial in learning new sounds in a second language, only new linguistic information consistent with the existing knowledge is often prioritized for consolidation. What remains unclear from the research is whether prior tonal knowledge from a native language (i.e., pitch contour signaling the Mandarin contour-tone system) influences an overnight consolidation of tone learning. The study adopts an overnight design, using Cantonese contour and level tones contrasting in pitch contour and height, for two perceptual learning experiments conducted separately on Mandarin and English-speaking novice learners of Cantonese. The first experiment found that Mandarin-speaking participants showed a stronger effect of consolidation in novel words contrasting in contour tones than in level tones, thanks to their prior knowledge of contour tones. The consolidation effect was predicted by rough estimates of deep-sleep length. Without prior knowledge of tones, English-speaking L2 learners in the second experiment showed an (unexpected) offline improvement for both contour and level tones. Overall, the findings suggest a preferential effect on overnight consolidation of contour tones when the cues contrasting L2-Cantonese tones are consistent with L1-Mandarin prior knowledge.
View
... Of course, PAM is mostly concerned with recognition of higher order phonological structures in perception, but it does not focus on how listeners exploit acoustic signals for matching representations with sensory material in a computational sense. When the acoustic cues for a specific phonological structure can be easily defined, e.g., voice-onset-time (VOT) for voiceless and voiced plosives, or formants and duration for vowels, psychoacoustic models of nonnative speech perception, e.g., Speech Learning Model (Flege, 1995;Flege & Bohn, 2021) can also generate testable predictions based on language-specific ways of acoustic signal processing. ...
View
... Bulgarian /ɛ ɔ/ are typically open-mid, while most varieties of Spanish tend to have closer mid vowels, [e o]. Research in bilingual speech has often been articulated in the framework of the Speech Learning Model (Flege, 1995;Flege & Bohn, 2021), which posits that bilinguals strive to differentiate their two languages' phonetic categories, which coexist in a shared mental phonetic space. Where two sounds are perceived as distinct, but are cognitively linked as similar, the learner may form a single composite -or compromise -category based on the combined distribution of the input; with time, this composite category can become narrower and assimilate to the Table 1 The Bulgarian stressed vowel system (adapted from Sabev, 2023, p. 2). ...
Contrast Maintenance in the Bulgarian and Judeo-Spanish Unstressed Vowels of Heritage Bilinguals
Article
  • Apr 2025
We present an acoustic-phonetic investigation of vowels in the spontaneous speech of heritage speakers of Judeo-Spanish (/ɛ a ɔ i u/) and Bulgarian (/ɛ a ɔ i ɤ u/), alongside monolingual Bulgarians' vowels. Formant frequencies and duration were analyzed for 9,498 vowel tokens. Linear mixed models revealed no spectral differences between Judeo-Spanish and Bulgarian stressed vowels, pointing to shared targets in the
View
Speech Enhancement Algorithms: A Systematic Literature Review
Article
  • May 2025
A growing and pressing need for Speech Enhancement Algorithms (SEAs) has emerged with the proliferation of hearing devices and mobile devices that aim to improve speech intelligibility without sacrificing speech quality. Recently, a tremendous number of studies have been conducted in the field of speech enhancement. This study aims to map the field of speech enhancement by conducting a systematic literature review to provide comprehensive details of recently proposed SEAs. This systematic review aims to highlight research trends in SEAs and direct researchers to the most important topics published between 2015 and 2024. It attempts to address seven key research questions related to this topic. Moreover, it covers articles available in five research databases that were selected in accordance with the PRISMA protocol. Different inclusion and exclusion criteria have been performed. Across the selected databases, 47 studies met the defined inclusion criteria. A detailed explanation of SEAs in the recent literature is provided, with existing SEAs studied in a comparative fashion along with the factors influencing the choice of one over the others. This review presents different criteria related to the approaches utilized for signal modeling, the different datasets employed, types of transform-based SEAs, and the effectiveness of different measurements, among other topics. This study presents a systematic review of SEAs along with existing challenges in this field.
View
View
Spanish rhotic acquisition in semester-long study-abroad and at-home contexts
Article
  • Apr 2025
The present study compares the L2 acquisition of Spanish rhotics (intervocalic taps /ɾ/ and trills /r/) by 16 semester-long study-abroad (SA) participants and 7 at-home (AH) peers using an experimental production task. Data collected at three different time points (pre-/mid-/post-semester) suggest an advantage of study-abroad in the acquisition of trills containing multiple occlusions. Participants at all proficiency levels showed significant movement towards multiple-occlusion trills throughout the semester. Intensity of contact with Spanish interacted with proficiency, enhancing gains at certain levels. Results indicate that segment duration can also serve as a useful phonemic contrast. While both the SA and AH groups significantly increased trill segment duration over time, only the SA group sustained the increase from mid- to post-semester. These findings have theoretical and pedagogical implications for the acquisition of L2 phonetic categories and phonemic distinctions that differ from the L1.
View
View
Impact of early bilingualism on perception and production of new sounds of a third language in young children
Article
  • Apr 2025
  • INT J BILINGUAL
Aims and objectives In this study, we examine the discrimination and the pronunciation of new Spanish and Mandarin phonemes in French–English bilingual children in first and second grades compared with French-speaking monolingual children of the same grades. Design/methodology/approach We recruited 18 French–English bilingual children and 27 French-speaking monolingual children in first or second grade in elementary schools. They were asked to carry out discrimination and word repetition tasks from audio files recorded by a Colombian Spanish speaker and a Chinese Mandarin speaker. Data and analysis For discrimination tasks, we calculated two measures: the percentage of success in the discrimination of two Spanish rhotics (/r/-/ɾ/) and the percentage of success in the discrimination of the set of Mandarin phonological contrasts (/p-p h /, /t-t h /, /k-k h /, /ʂ-ɕ/, /ts-ts h /, /tʂ-tʂ h / and /tɕ-tɕ h /). For word repetition tasks, pronunciation was assessed for accuracy and foreign accent by native Spanish and Mandarin speakers. We completed our analyses with measures of voice onset time (VOT) values of Mandarin stops in word initial position (in milliseconds). Findings and conclusions The results of our study showed that the members of the bilingual group were neither advantaged nor disadvantaged compared with the monolingual group. Our findings suggest that specific exposure to the characteristics intrinsic to an L3 is necessary for mastering new sounds in an L3. Originality This study examines whether early bilingualism facilitates the perception and the pronunciation of new sounds in an L3 in young bilingual (French–English) children. Few studies have examined this issue in this age group, that is, 6- to 9-year-old children. In addition, unlike other studies that analyzed oral production in speech, we chose to focus on the pronunciation of sounds in isolated words. Significance Our study highlights that no significant differences between bilingual and monolingual children were revealed. These results contribute to an understanding of whether there are bilingual advantages in L3 phonological acquisition.
View
Examination of Perceptual Reorganization for Nonnative Speech Contrasts: Zulu Click Discrimination by English-Speaking Adults and Infants
Article
Full-text available
  • Aug 1988
The language environment modifies the speech perception abilities found in early development. In particular, adults have difficulty perceiving many nonnative contrasts that young infants discriminate. The underlying perceptual reorganization apparently occurs by 10–12 months. According to one view, it depends on experiential effects on psychoacoustic mechanisms. Alternatively, phonological development has been held responsible, with perception influenced by whether the nonnative sounds occur allophonically in the native language. We hypothesized that a phonemic process appears around 10–12 months that assimilates speech sounds to native categories whenever possible; otherwise, they are perceived in auditory or phonetic (articulatory) terms. We tested this with English-speaking listeners by using Zulu click contrasts. Adults discriminated the click contrasts; performance on the most difficult (80% correct) was not diminished even when the most obvious acoustic difference was eliminated. Infants showed good discrimination of the acoustically modified contrast even by 12–24 months. Together with earlier reports of developmental change in perception of nonnative contrasts, these findings support a phonological explanation of language-specific reorganization in speech perception. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
View
View
View
View
View
View
View
View
Factors Affecting the Production of Word-initial Consonants in a Second Language
Chapter
  • Jan 1996
This volume corrects the relative neglect in Second Language Acquisition studies of the quantitative study of language variation and provides insights into such issues as language transfer, acquisition through exposure, language universals, learner’s age and so forth. These studies bolster the idea that a full account of SLA development (and, hence, a “theory of SLA”) must be built on not only detailed accounts of interlanguage data but also on a wide appeal to factors which govern the psycholinguistic bases of SLA. An important addition to the volume is a comprehensive guide to both the DOS and Macintosh versions of the VARBRUL statistical program used by variationists.
View
View