Research ArticleLexical representations can rapidly be updated in the early stages of second-language word learning
Introduction
When learning a second language (L2), one is faced with sounds that are not part of the native phonological inventory. Since, for late L2 learners, the perceptual system is attuned to the sounds and sound distinctions that are relevant in the native language (L1; Best & Tyler, 2007; Kuhl et al., 1992, Polka and Werker, 1994, Werker and Tees, 1984, Escudero, 2005, Flege, 1995), how the non-native sounds will be perceived is largely determined by the learner’s L1 phonological system. Especially problematic for L2 learning are cases in which two non-native sounds are perceptually similar to one and the same L1 category. This L1-L2 mapping configuration makes it particularly difficult to learn to perceptually distinguish the two L2 sounds, as both are associated to a common L1 counterpart. Well known examples of such “difficult” L2 contrasts are the English distinctions between /l/ and /r/ for native speakers of Japanese (Bradlow et al., 1999, Goto, 1971, Ingvalson et al., 2011), between /i/ and /ɪ/ for native speakers of Italian and Spanish (e.g., Casillas, 2015, Cebrian, 2006, Flege et al., 1995) and between /ɛ/ and /æ/ for native speakers of Dutch and German (e.g., Bohn and Flege, 1990, Bohn and Flege, 1992, Broersma, 2012, Eger and Reinisch, 2019a, Eger and Reinisch, 2019b, Escudero et al., 2008, Llompart, 2021a, Llompart, 2021b, Llompart and Reinisch, 2017, Llompart and Reinisch, 2019a, Llompart and Reinisch, 2019b, Llompart and Reinisch, 2020).
In order to fully master these difficult L2 contrasts, learners need to reach two interconnected milestones. Firstly, they have to overcome the perceptual difficulties caused by the contrast and manage to establish separate phonetic categories for the two sounds. Secondly, learners also need to be able to link these categories to the L2 words that should contain them. In more technical terms, they need to encode the two non-native phonetic categories into the phonological forms of the corresponding L2 lexical representations. For instance, for the English /ɛ/-/æ/ contrast, which is the object of the present study, this means that /ɛ/ has to be encoded into the representations of words like bet and lemon, and /æ/ into those of bat and dragon–and not the other way around. This process has been referred to in recent studies as phonolexical encoding, or simply lexical encoding (Cook and Gor, 2015, Cook, Pandža, Lancaster, & Gor, 2016, Darcy et al., 2013, Darcy and Thomas, 2019, Llompart and Reinisch, 2019b, Llompart and Reinisch, 2020, Hayes-Harb and Barrios, 2019). In the present study, we assess how training that focuses on the phonological form of words affects the lexical encoding of phonetic categories corresponding to difficult non-native contrasts. This is done by means of a word learning paradigm in which novel L2 minimal pairs only differing in the sounds of a difficult L2 contrast are presented. More specifically, our main research question is whether the point in time at which form-focused training is administered (i.e., very first presentation vs. once lexical representations are already in place) determines the robustness of learners’ resulting lexical encoding.
When examining the relationship between the establishment of L2 phonetic categories and their encoding into lexical items, a recurrent finding has been that learners have substantial difficulties with the latter above and beyond the challenges that are known to relate to the former (Amengual, 2016, Darcy et al., 2013, Díaz et al., 2012, Llompart and Reinisch, 2019b). These lexical difficulties have major repercussions on L2 communication, since a robust lexical encoding of potentially challenging L2 sounds is essential to avoid confusions with minimal pairs when these cannot be disambiguated by the context (Broersma, 2012, Ota et al., 2009) and to prevent spurious lexical activation and competition in spoken word recognition (Broersma and Cutler, 2008, Broersma, 2012, Cook and Gor, 2015, Llompart and Reinisch, 2019b, Sebastián-Gallés et al., 2005, Weber and Cutler, 2004).
It is widely known that the phonetic identification of difficult L2 sounds improves with time and L2 exposure (e.g., Bohn and Flege, 1990, Flege et al., 1997, Eger and Reinisch, 2019a, Llompart, 2021b, Simonchyk and Darcy, 2017). Similarly, lexical encoding also becomes increasingly robust as a function of L2 proficiency (e.g., Darcy et al., 2012, Llompart, 2021a, Llompart, 2021b, Simonchyk and Darcy, 2017, Simonchyk and Darcy, 2018). In fact, a relationship between the two at the level of the individual learner has been found in several studies, at least for learners of intermediate proficiency levels (Darcy and Holliday, 2019, Llompart, 2021b, Silbert et al., 2015, Simonchyk and Darcy, 2017). However, very little is known about the process by which the phonological form of existing L2 lexical representations is updated to reflect learners’ improvement, as well as about how such updating is fostered or constrained.
Darcy and Holliday (2019) set out to address this issue by assessing whether Mandarin learners of Korean had encoded the sounds of a difficult L2 contrast (/o/-/ʌ/) more robustly into recently learned words (“new words”) than in words that they had long known (“old words”). This distinction was made in accordance with the words’ appearance in textbooks for learning Korean as an L2. The rationale behind this was that, if L2 learners exhibited a less robust encoding for old words, this would suggest that the updating of L2 lexical representations is costly and not always possible. If, on the contrary, encoding for new and old words did not differ, this would suggest that lexical representations for old words had been updated to match the encoding of newer words and thus reflect learners’ current phonological knowledge. In order to test these hypotheses, Mandarin learners of Korean were asked to perform a lexical decision task containing nonwords created by swapping the two critical vowels (e.g., [ʨʌnjʌk] “dinner” produced as *[ʨonjʌk]). Results failed to show significant differences in nonword rejection accuracy between the new and old words, which in principle suggests that old words’ representations had been updated. However, as the authors acknowledge, this null result may be due to a lack of statistical power because of the reduced sample size.
The study by Darcy and Holliday emphasizes that any attempt to assess the updating of L2 lexical representations using real words is bound to face multiple obstacles. In the first place, experience with specific L2 words may differ largely between individuals and is very hard to quantify. For example, determining when a word was first heard, or how many times a word has been heard is an almost impossible enterprise. Secondly, without detailed longitudinal data, it is virtually impossible to establish how robustly the sounds in difficult non-native contrasts were encoded into specific L2 words at particular points in time and how that encoding changed over time (see Flege & Bohn, 2020).
A workaround to these limitations is resorting to novel word learning paradigms as an alternative to investigate the encoding of difficult phonological contrasts into L2 words in real time (Escudero et al., 2008, Hayes-Harb and Masuda, 2008, Llompart, 2021b, Llompart and Reinisch, 2019b). In novel L2 word learning, participants who already have experience with the language in question learn to associate pairs of partially-overlapping nonwords that are phonologically plausible in the L2 to pictures of novel objects through an intensive training session including feedback on their responses. After training, their identification accuracy for the novel L2 words is assessed. The main advantages of novel word learning are that i) experimental outcomes are not dependent on learners’ previous experience with the words to be learned, and ii) it targets the very earliest (i.e., initial) stages of word learning, which are usually far gone for most known words of the non-native language.
A potential disadvantage of novel word learning paradigms is, nonetheless, that unless starkly longitudinal designs are implemented, learners’ phonetic perception of the critical contrast at the moment of learning the words will be constant across all words learned (vs. the assumption for real words in Darcy & Holliday, 2019). However, previous studies have shown that the encoding of L2 phonetic categories into novel lexical items is highly dependent on the information that is provided during word learning. In short, L2 learners appear not to be able to establish a phonological contrast in novel words between confusable L2 sounds, like /ɛ/ and /æ/ for Dutch and German speakers, unless their attention is directed towards the critical sounds during the word learning process (Escudero et al., 2008, Llompart, 2021b, Llompart and Reinisch, 2019b). The critical attention allocation can be achieved either through exposure to additional contrastive information (Escudero et al., 2008, Llompart and Reinisch, 2017), or by making the stimuli and learning situation directly point towards the existence of a distinction, like when listeners are trained to learn novel minimal pairs and are explicitly asked to choose between the members of the pairs as part of the training procedure (Llompart & Reinisch, 2020).
The latter scenario is a form of what has been called “phonological specificity training”, a training paradigm used to teach minimal pairs that is designed to enhance the distinctiveness of words’ phonological representations. This is done by visually presenting the referents of the target words together with the referents of their minimal-pair competitors and prompting listeners to decide between the two based on the auditory input they receive (Janssen et al., 2015, van de Ven et al., 2019, van Goch et al., 2014). Phonological specificity training has been shown to have positive effects on phonological awareness and vocabulary learning in children learning their L1 and L2 (Janssen et al., 2015, van Goch et al., 2014), as well as in adolescents learning an L2 (van de Ven et al., 2019).
Importantly, if the introduction of phonological specificity training to the procedure is set to vary in time, a novel L2 word learning paradigm (Escudero et al., 2008, Llompart, 2021b, Llompart and Reinisch, 2019b) designed to encompass multiple training phases can lead to new insights on the updating of lexical representations containing confusable L2 phonological contrasts. This is exactly what we did in the present study. German learners of English were taught 5 minimal pairs of novel English words, in which one member of the pair had /ɛ/ as first vowel and the other had /æ/ (e.g., tendek-tandek), by means of a novel word learning paradigm including two training phases. The training phases differed in whether they involved phonological-form focus through phonological specificity training or not and, crucially, the moment at which learners were exposed to phonological specificity training was manipulated across groups (form-focused training was presented first or second). We report two experiments that addressed the issue of the timing of specificity training during word learning. The two experiments differed in the type of input that learners received in the training phases. In Experiment 1, learners were only exposed to one and the same native speaker of English in all training and test phases. In Experiment 2, comparable groups of learners were exposed not only to the same native speaker of English but also to novel-word productions by a German-accented learner of English. The rationale for including same-L1-accented tokens in Experiment 2 was that, in most non-immersion classroom situations, learners are faced not only with native input but also with accented input from their peers or even the teacher. This issue will be discussed in detail the introduction to this second experiment.
Section snippets
Experiment 1
In Experiment 1, German learners of English were taught 5 /ɛ/-/æ/ minimal pairs by means of a two-day (henceforth referred to as “day 1” and “day 2”) novel word learning paradigm involving two training phases (i.e., one on each day) and three test phases at different points in time assessing their recognition of the novel words. In one training phase, the referents of the target words were presented on the screen together with their competitors (i.e., phonological specificity training), whereas
Experiment 2
Experiment 1 showed that phonological specificity training is highly beneficial in early novel word learning of items containing confusable L2 categories and that L2 lexical representations can be updated to reflect the effects of such training. Nevertheless, Experiment 1 created a learning situation that may not be replicated in real word learning in an L2, especially if learners are in a non-immersion setting like the German learners of English tested here. In this first experiment, as in
General discussion
The two experiments in the present study assessed the effect of (phonological) form-focused training in a two-day novel L2 word learning paradigm in which German learners of English learned minimal pairs only differing in a difficult phonological contrast (/ɛ/-/æ/). In particular, we addressed the issue of how differences in the timing at which phonological specificity training is undergone (day 1 vs. day 2) impacts the lexical encoding of /ɛ/ and /æ/ in the novel words. The main aim of the
CRediT authorship contribution statement
Miquel Llompart: Conceptualization, Methodology, Software, Investigation, Formal analysis, Visualization, Data curation, Writing - original draft, Writing - review & editing. Eva Reinisch: Conceptualization, Methodology, Investigation, Writing - original draft, Writing - review & editing.
Acknowledgments
This project was conducted while the second author was at Ludwig Maximilian University in Munich, Germany. The project was funded by a grant of the German Research Foundation (RE3047/1-1) to the second author. The first author was funded by an Alexander von Humboldt Professorship (ID-1195918) awarded to Ewa Dąbrowska, Chair of Language and Cognition at the Department of English and American Studies of Friedrich Alexander University Erlangen-Nuremberg. We would like to thank Nicole Benker, David
References (77)
- et al.
Phantom word activation in L2
System
(2008) Experience and the use of non-native duration in L2 vowel categorization
Journal of Phonetics
(2006)- et al.
Acquiring an artificial lexicon: Segment type and order information in early lexical entries
Journal of Memory and Language
(2006) - et al.
Individual differences in late bilinguals’ L2 phonological processes: From acoustic-phonetic analysis to lexical access
Learning and Individual Differences
(2012) - et al.
Novel second-language words and asymmetric lexical access
Journal of Phonetics
(2008) - et al.
Effects of experience on non-native speakers’ production and perception of English vowels
Journal of Phonetics
(1997) - et al.
Lexical competition and the acquisition of novel words
Cognition
(2003) Auditory perception by normal Japanese adults of the sounds “l” and “r”
Neuropsychologia
(1971)- et al.
Predicting native English-like performance by native Japanese speakers
Journal of Phonetics
(2011) - et al.
The KEY to the ROCK: Near-homophony in nonnative visual word recognition
Cognition
(2009)
Does listening to non-native speech impair speech perception?
Journal of Memory and Language
The influence of initial exposure on lexical representation: Comparing early and simultaneous bilinguals
Journal of Memory and Language
Non-native phonemic discrimination, phonological short term memory, and word learning
Journal of Phonetics
Lexical competition in non-native spoken-word recognition
Journal of Memory and Language
Cross-language speech perception: Evidence for perceptual reorganization during the first year of life
Infant Behavior and Development
Assessing the effects of phonetic training on L2 sound perception and production
The perception of language-specific phonetic categories does not guarantee accurate phonological representations in the lexicon of early bilinguals
Applied Psycholinguistics
The impact of L2 dialect on learning French vowels: Native English speakers learning Québécois and European French
Canadian Modern Language Review
Effects of acoustic variability on second language vocabulary learning
Studies in Second Language Acquisition
Effects of variability in fundamental frequency on L2 vocabulary learning: A comparison between learners who do and do not speak a tone language
Studies in Second Language Acquisition
Fitting linear mixed-effects models using lme4
Journal of Statistical Software
Interlingual identification and the role of foreign language experience in L2 vowel perception
Applied Psycholinguistics
The production of new and similar vowels by adult German learners of English
Studies in Second Language Acquisition
Training Japanese listeners to identify English /r/ and /l/: Long-term retention of learning in perception and production
Perception & Psychophysics
Increased lexical activation and reduced competition in second-language listening
Language and Cognitive Processes
Production and perception of the /i/-/ɪ/ vowel contrast: The case of L2- dominant early learners of English
Phonetica
Lexical access in L2: Representational deficit or processing constraint?
The Mental Lexicon
Fuzzy nonnative phonolexical representations lead to fuzzy form-to-meaning mappings
Frontiers in Psychology
Impossible to _gnore: Word-form inconsistency slows preschool children’s word-learning
Language Learning and Development
Asymmetric lexical access and fuzzy lexical representations in second language learners
The Mental Lexicon
Direct mapping of acoustics to phonology: On the lexical encoding of front rounded vowels in L1 English-L2 French acquisition
Second Language Research
When blue is a disyllabic word: Perceptual epenthesis in the mental lexicon of second language learners
Bilingualism: Language and Cognition
The formants of monophthong vowels in Standard Southern British English pronunciation
Journal of the International Phonetic Association
Sleep-associated changes in the mental representation of spoken words
Psychological Science
The impact of one’s own voice and production skills on word recognition in a second language
Journal of Experimental Psychology: Learning, Memory, and Cognition
The role of acoustic cues and listener proficiency in the perception of accent in nonnative sounds
Studies in Second Language Acquisition
Cited by (6)
Spelling provides a precise (but sometimes misplaced) phonological target. Orthography and acoustic variability in second language word learning
2022, Journal of PhoneticsCitation Excerpt :As predicted, for L2 English words learned with both audio and orthographic input, our L1 French participants produced vowels that were both more precise, more tightly clustered around a target (less disperse, smaller DS), and misplaced (with respect to the English learning input) in the vowel space, with loci corresponding to L1 French vowels (smaller ED). A new word added to the L2 lexicon may have a phonolexical representation that is less precise, “fuzzy”, vague or “malleable” (Hayes-Harb & Masuda, 2008; Darcy, Daidone, & Kojima, 2013; Cook, Pandža, Lancaster, & Gor, 2016; Llompart, 2021; Llompart & Reinisch, 2021). In addition, there is evidence that certain types of lexical information, such as phonological information, may be more heavily weighted in the L2 lexicon than other types of information, such as semantic information (for a review, see Cook et al., 2016).
Whose input matters?: The influence of socially-differentiated input sources in adult Lx phonetic learning
2022, Journal of Second Language PronunciationAptitude and experience as predictors of grammatical proficiency in adult Greek-English bilinguals
2022, Frontiers in Psychology