Session 3A

Attention has been shown to be a pre-requisite for learning to take place (Schmidt, 2001; Leow, 2012) and orienting attention through explicit instruction has been found to facilitate learning during high variability phonetic training (HVPT) procedures (Pederson & Guion, 2010; Nozawa, 2015; Alves & Luchini, 2016). This study further explored the effects of two HVPT methods (Identification (ID) or Categorial discrimination (DIS)) on attended and unattended target sounds. The training regimes aimed at improving the perception of English stop consonants (/p/, /t/, /k/, /b, /d, /g/) and five vowels (/i/, /?/, /æ/, /?/, /??/) by Spanish/Catalan bilingual learners of English. Participants were divided into four experimental groups (N=20) and a control group (N=16) and were perceptually tested before and after a five-week training period. The experimental groups differed either in terms of training method (ID or DIS) or focus of training (consonants or vowels), resulting in four different groups. Thus, two groups were told to attend to vowels and two groups were told to attend to consonants embedded in the exact same set of stimuli. All groups were tested on both consonants and vowels, thus assessing the effects of the two perceptual training methods on attended and unattended segments. Moreover, retention of knowledge was tested two months after the completion of the training regime. Findings revealed that all experimental groups significantly outperformed the control groups in their identification of attended target sounds, and were able to retain the knowledge, confirming the efficacy of both training methodologies (Flege, 1995; Nozawa, 2015) and the important role that attention orienting plays in L2 phonetic learning (Pederson & Guion, 2010). However, the DIS trainees showed significant improvement on the unattended target sounds as well as on the attended ones, suggesting that balanced exposure to the stimuli was sufficient to promote learning during a categorial discrimination training regime. These findings will be discussed alongside with the pedagogical implications of each perceptual training method as EFL learning tools.

High-variability phonetic training (HVPT) has been shown to lead to robust gains in learners’ perception and production of L2 vowel contrasts [1]. However, such gains might be influenced by learners’ individual differences [2], as well as by training conditions, i.e. number of training sessions, number of trials and breaks within sessions [3-5], number of vowel contrasts trained, or type of training materials, such as the lexical status of the training stimuli [6]. For example, nonword-based training has been shown to be superior to word-based training as it is phonetically oriented and avoids lexical interference during training [6]. However, no research to date has examined the potential impact of training materials (nonword vs. word training stimuli) on the learning curves of those who benefit from HVPT, which may provide an insight into learning mechanisms and the optimal design of vowel training programmes. The present study therefore examined the extent to which HVPT improved adult L2 learners' perception and production of L2 vowels, and whether gains and individual learning curves differed as a function of lexical status of training materials. A group of Catalan-Spanish learners of English (N=65) participated in four 40-minute sessions of HVPT targeting a difficult L2 vowel contrast (/æ/-/?/) that perceptually assimilates to a single native category /a/. Participants were randomly assigned to one of two stimuli training conditions (nonwords or words) and trained through AX discrimination, identification and immediate repetition tasks. Trainees completed 352 trials per session (352 x 4 sessions=1408). Perception gains (discrimination accuracy) were assessed through a categorical ABX discrimination task, while production gains (spectral distance scores between learners’ and native speakers’ vowel productions, for each vowel and trial) were assessed through a delayed word repetition task. An untrained group (N=15) served as controls. In order to assess the rate at which trainees improved across training sessions, we calculated the linear slope of each individual’s learning curve between pre-test and post-test, and between training sessions. Preliminary analyses revealed that HVPT led to overall gains in L2 perception, irrespective of training stimuli, and more limited gains in L2 production, being nonword-based training more efficient than word-based training. In the non-word training condition, largest gains were found in the first two training sessions, and then the size of gains reduced progressively. From the point of view of training efficacy, this finding suggests that training with non-words may not only be superior to training with words, but may also lead to faster gains in L2 perception and production. References [1] Iverson, P., Pinet, M., & Evans, B. G. (2012). Auditory training for experienced and inexperienced second-language learners: Native French speakers learning English Applied Psycholinguistics, 33 (01), 145-160. [2] Perrachione, T. K., Lee, J., Ha, L. Y., & Wong, P. C. (2011). Learning a novel phonological contrast depends on interactions between individual differences and training paradigm design. The Journal of the Acoustical Society of America, 130(1), 461-472 [3] Aberg, K.C., Tartaglia, E.M., Herzog, M.H. (2009). Perceptual learning with Chevrons requires a minimal number of trials, transfers to untrained directions, but does not require sleep. Vision Res 49, 2087–2094. [4] Wright, B.A., Sabin, A.T. (2007). Perceptual learning: how much daily training is enough? Exp Brain Res 180, 727–736. [5] Molloy, K., Moore, D.R., Sohoglu, E., Amitay, S. (2012). Less Is More: Latent Learning Is Maximized by Shorter Training Sessions in Auditory Perceptual Learning. PLoS ONE 7(5), e36929. [6]Thomson, R, Derwing, T. (2016). Is phonemic training using nonsense or real words more effective? Proceedings of the 7th Pronunciation in Second Language Leanring and Teaching Conference, 88-97.

While some L2 learners might wish to acquire native-like pronunciation, most of them strive to attain comprehensible speech. Thus, the current focus of L2 speech development and teaching is not on native-likeness, but on comprehensibility, that is, listeners' ease of understanding (Saito et al., 2017). L2 speech researchers have, therefore, tried to identify those pronunciation features that are important for achieving successful comprehensibility regardless of accentedness (Isaacs & Trofimovich, 2012). Research shows that while listeners’ accent judgements are mainly influenced by pronunciation accuracy, comprehensibility judgements are in addition linked to grammatical accuracy and lexical richness (Isaacs & Trofimovich, 2012). Most studies to date have examined comprehensibility of L2 speech through native listeners’ judgements, but little is known about L2 learners' perception of their own speech and the factors that might influence their judgements. Inaccurate speech self-assessment might be indicative of learners' failure at noticing pronunciation and lexico-grammatical errors that affect their comprehensibility and could negatively impact L2 speech development. The few existing studies that have investigated L2 speech self-assessment (Strachan et al., 2019; Trofimovich et al., 2016) have found learners at the low end of the comprehensibility and accentedness scales to overestimate and those at the high end to underestimate their own speech. The present study extends this line of research by examining L2 learners' self-assessments of their own speech for comprehensibility and the extent to which these matched the ratings of non-native peers and native listeners. It also investigates learners’ perception of task performance (i.e. overall performance, anxiety, difficulty and cognitive effort) and how it might relate to comprehensibility and accentedness self-assessments. Speech samples were elicited and recorded from 75 undergraduate learners of L2 English performing a picture-based story telling task (Derwing et al., 2009). They were asked to evaluate their task performance immediately after the task and, on a different day, to self-assess their own speech by listening to their full speaking performance for comprehensibility and accentedness through nine-point scales. They were also asked to qualitatively describe the speech dimensions they were paying attention to during the rating task. Peer undergrad non-native listeners (N=30) as well as native English listeners (N=7) evaluated the learners’ full productions (M=76 sec, SD=38 sec) for the same dimensions in the same way. The results indicate, in line with previous research, that learners’ self-ratings of comprehensibility and accentedness did not accurately match the ratings of native and non-native listeners. Learners either overestimated or underestimated their oral performance and linked their comprehensibility to both pronunciation and lexico-grammar dimensions and degree of accentedness to various aspects of L2 phonology. While comprehensibility ratings were related to perception of overall task performance, anxiety and effort, accentedness ratings only related to amount of effort in task performance. References Derwing, T.M., Munro, M.J., Thomson, R.I. and Rossiter, M.J. (2009). The relationship between L1 fluency and L2 fluency development. Studies in Second Language Acquisition, 31(4), 533-557. Isaacs, T., & Trofimovich, P. (2012). Deconstructing comprehensibility: Identifying the linguistic influences on listeners’ L2 comprehensibility ratings. Studies in Second Language Acquisition, 34, 475-505. Saito, K., Trofimovich, P., & Isaacs, T. (2017). Using listener judgements to investigate linguistic influences on L2 comprehensibility and accentedness: A validation and generalization study. Applied Linguistics, 38, 439-462. Strachan, L., Kennedy, S. and Trofimovich, P. (2019). Second language speakers’ awareness of their own comprehensibility. Journal of Second Language Pronunciation, published online 14th March. Trofimovich, P., Isaacs, T., Kennedy, S., Saito, K. and Crowther, D. (2016). Flawed self-assessment: Investigating self- and other-perception of second language speech. Bilingualism: Language and Cognition, 19 (1), 122-140.

Models of second language (L2) speech perception attempt to describe the ways in which target phones are assimilated to (i.e., perceived in terms of) L1 sound categories and their consequences for L2 perception and production. The Perceptual Assimilation Model-L2 (PAM-L2) (Best and Tyler, 2007) makes predictions about discrimination of L2 phones based on the degree to which pairs of target phones are assimilated to one or more L1 categories. The Natural Referent Vowel (NRV) framework (Polka and Bohn, 2011) tries to account for directional asymmetries observed in vowel perception. Vowels that are more peripheral in the acoustic space (in terms of first and second formant structure) act as perceptual anchors resulting in greater discrimination accuracy if the more peripheral vowel is presented second than if the order is the reverse. Another factor that may affect speech perception is the role of lexical representations (Ganong, 1980). For instance, L2 learners have been found to be more successful discriminating challenging L2 phones in real words than in nonwords (Mora, 2005). This paper investigates the effect of perceptual and lexical factors on discrimination of L2 vowel pairs by comparing the discrimination of L2 vowels presented in real and in nonsense words. In addition, the choice of L2 vowel pairs used in the discrimination test allows the evaluation of the models’ predictions based on cross-linguistic perceived similarity and vowel peripherality. Forty-four L1 Spanish L2 English speakers performed a perceptual assimilation task, in which they identified L2 vowels in terms of L1 categories, and two L2 vowel discrimination tasks. The discrimination tasks presented the members of each pair in the two possible orders (e.g., /??/-/??/, /??/-/??/), and stimuli were embedded in real and non-sense words. Results showed that discrimination accuracy was greater in real words than in non-sense words, highlighting the influence of lexical status in L2 vowel perception. Perceptual asymmetries emerged generally in accordance with the predictions of the NRV framework, mirroring previous findings in infant and adult perception. The results for three of the four vowel pairs tested were in agreement with the PAM-L2 predictions, but discrimination accuracy was not always explained by crosslinguistic mapping relations. It is possible that other factors are at play regarding the discrimination of specific contrasts such as the relative weighting of different acoustic cues or individual differences in the perceived similarity between L1 and L2 vowels. Best, C.T., Tyler, M.D. 2007. Nonnative and second-language speech perception: Commonalities and complementarities. In Bohn, O-S, Munro, M. J. (eds), Language Experience in Second Language Speech Learning. Amsterdam: John Benjamins, 13–34. Ganong, W. F. 1980. Phonetic categorization in auditory word perception. Journal of Experimental Psychology: Human Perception and Performance, 6(1):110-125. Mora, J. C. 2005. Lexical knowledge effects on the discrimination of non-native phonemic contrasts in words and nonwords by Spanish/Catalan bilingual learners of English. In Proceedings of ISCA Workshop on Plasticity in Speech Perception, 43-46. Polka, L., Bohn, O.-S. 2011. Natural Referent Vowel (NRV) framework: An emerging view of early phonetic development. Journal of Phonetics, 39(4), 467-478. Tyler, M. D., Best, C. T., Faber, A., & Levitt, A. G. 2014. Perceptual assimilation and discrimination of non-native vowel contrasts. Phonetica, 71(1), 4–21.