Session 1B

A growing number of researchers have investigated the role of emotional and attitudinal/motivational variables in foreign language learning and performance in a wide variety of contexts (Dewaele et al., 2019; Li, 2019; MacIntyre et al., 2019; Saito et al., 2018) but no work to date has investigated the relationship between the same learners’ classroom emotions, attitudes and motivation in their first language (L1) and in their two different foreign languages, nor their combined effect on exam performance in the L1, L2 and L3. Such work is important because it would allow some degree of generalization about the dynamic relationship between classroom emotions, attitudes and motivation in language learning in general. Data were collected from 113 secondary school pupils in an Italian immersion school in Istanbul, Turkey. In addition to Turkish L1 classes, pupils had classes in Italian L2 and English L3. Non-parametric statistical analyses revealed that levels of Enjoyment and Attitudes/Motivation were broadly similar in the three languages but that levels of Anxiety were significantly different, with increased Anxiety in the foreign languages. Further analyses showed that levels of Enjoyment and Attitudes/Motivation (but not Anxiety) were significantly correlated across languages (small to medium effect sizes) suggesting that some pupils share a love of languages in general. A significant positive relationship emerged between Enjoyment and Attitude/Motivation within each language (medium to large effect sizes). Anxiety was found to be negatively linked to Enjoyment and Attitude/Motivation in Italian and English (small to medium effect sizes). Analyses showed that age, gender and degree of multilingualism had only limited and scattered effects on the dependent variables in the three languages. Finally, pupils with high Anxiety were found to have lower exam scores in the L1, L2 and L3 while Enjoyment and Attitude/Motivation were positively related to exam results in Italian L2 and English L3. It thus seems that the web of relationships between classroom emotions and Attitudes/Motivation is not fundamentally different in the L1, L2 and L3 – with some limited variation- and that Anxiety has a wider-ranging negative effect on exam performance than Enjoyment and Attitude/Motivation. The present study confirms the consistent finding in previous research that positive and negative emotions are not unidimensional (Dewaele & MacIntyre, 2014) but that they have unique effects on language learning and performance, and that they are linked to more stable psychological constructs such as language attitudes and motivation, as well as the learning context. The pedagogical implication of the present study is that language teachers should strive to find the right emotional temperature in the classroom in order to strengthen learners’ Attitude/Motivation and ultimately boost their progress and performance in all languages. References Dewaele, J.-M., Chen, X., Padilla, A.M., & Lake, J. (2019). The flowering of positive psychology in foreign language teaching and acquisition research. Frontiers in Psychology, Language Sciences, 10, Doi:10.3389/fpsyg.2019.02128 Dewaele, J.-M., & MacIntyre, P. D. (2014). The two faces of Janus? Anxiety and enjoyment in the foreign language classroom. Studies in Second Language Learning and Teaching, 4, 237-274. Li, C. (2019). A positive psychology perspective on Chinese EFL pupils’ trait emotional intelligence, foreign language enjoyment and EFL learning achievement. Journal of Multilingual and Multicultural Development. doi: 10.1080/01434632.2019.1614187 MacIntyre, P. D., Dewaele, J.-M., Macmillan, N., & Li, C. (2019). The emotional underpinnings of Gardner’s attitudes and motivation test battery. In A. Al-Hoorie & P. D. McIntyre (Eds.), Contemporary language motivation theory: 60 years since Gardner and Lambert (1959). Bristol: Multilingual Matters, pp. 57-79. Saito, K., Dewaele, J.-M., Abe, M., & In’nami, Y. (2018). Motivation, emotion, language experience, and second language comprehensibility development in classroom settings: A cross-sectional and longitudinal study. Language Learning, 68, 1-35.

Theoretical framework The question of cognition in second language (L2) acquisition later in life is of importance inasmuch as language learning is largely mediated by domain-general cognitive capacities, such as working memory (e.g. Linck et al., 2014). A number of these capacities, however, have been shown to decline across the lifespan – although cognitive functions that increase with age have also been found (e.g. Berthele & Kaiser, 2014). At the same time, there is a pronounced degree of variability amongst older adults, and this variation has been shown to modulate cognitive performance (e.g. Christensen, 2001). Despite the well-documented cognition-language interface, much remains to be explored about the way individual differences in cognitive capacities affect individual L2 learning trajectories in the third age and to what extent this interaction may be mediated by demographic and socio-affective factors. Research questions The main goal of this micro-development study is to analyze the L2 trajectories of older German-speaking adults (aged 65+) in Switzerland who took part in an intensive 7-month Spanish course for beginners. Some of the key questions we ask the models are (1) when and why L2 development is statistically significantly increasing (or decreasing), and (2) whether our set of predictors (age, cognitive fitness, bilingualism, education) has a significant effect on the trajectories under investigation. Methods We made use of generalized additive mixed-effects regression models (GAMMs) to quantify linear and non-linear learner trajectories as well as any predictors thereof. Our data set contained 28 trajectories, each of them represented by 30 measurements taken at equal intervals over 7 months, amounting to 818 data points per test. Participants were assessed on a range of behavioral (e.g. working memory), L2 (receptive and productive skills) and demographic measures (education, age, previous foreign language experience), with the aim to identify factors that facilitate L2 learning and help explain inter- and intra-individual variation. Results Our models showed that there was significant (linear as well as non-linear) increase across all seven measures of L2 proficiency. In addition, L2 development could be shown to vary significantly between and within individuals in each of the seven measures. This was also confirmed by a GAMM-based visualization of periods of significant L2 growth on both the group level and the subject level: individual learner trajectories did not show the mean pattern of the group, as periods of significant growth differed between learners with respect to occurrence in the development, duration, degree of (non)linearity, height and steepness (see Fig. 1). In addition, the models confirmed that cognitive capacities, in particular working memory, predicted both linear and non-linear development in six of the seven L2 measures. Aside from the case of one measure of general L2 proficiency (C-test), the effect persisted even after including control factors such as education and prior foreign language experience in the models. Again, the influence of cognition on L2 development varied significantly between and within individuals. Innovation Data from such an iterated investigation will make a substantial contribution to research on late L2 acquisition as well as on cognitive aspects of healthy and active aging. Our research will also be crucial in the establishment of learner profiles and, subsequently, will inform adult educators about the development of appropriate teaching materials as well as the design of individualized language training.

Ever since the advent of the field, most SLA research has focused on the early developmental stages. For a variety of reasons, however, there is increasing interest now in relatively neglected ‘high-level proficiency’ (Hyltenstam, 2016, 2018; Malovrh & Benati, 2018). One of the earliest studies of advanced learners was that by Coppieters (1987). Based on a study of the grammaticality judgements and underlying intuitions of English-speaking learners of French, Coppieters concluded that the grammatical intuitions of very advanced non-native speakers (NNSs) of a language may diverge from those of native speakers (NSs), even about rules for which the NNSs’ performance passes as native-like. In part 1 of this presentation, we briefly report a study comparing the spontaneous judgements and underlying intuitions of 20 Spanish NSs and 41 very advanced English-speaking learners of Spanish on a noun gender-assignment task. In our study, gender assignment was assessed twice, speeded and unspeeded, using recordings of 100 phonologically and morphologically plausible non-words in Spanish (alicud, caspimbrón, manumbre, tolacín, etc.) spoken by a female NS of Spanish and simultaneously shown on a computer screen. The nouns were ones on whose gender at least 80% of NSs had agreed in a pilot study. Of those, 74 were classified as masculine, and 26 as feminine. In the first, speeded condition, designed to tap implicit knowledge, participants had three seconds to decide on each noun’s gender, and to indicate their decision by saying ‘masculine’ or ‘feminine’. In a subsequent informal interview with the researcher, they completed the second, unspeeded version of the same task, designed to tap any explicit knowledge they might have, and explained the perceived source of their intuitions about gender assignment, subsequently classified by the researchers as feel judgement, word ending, lexical cue, semantic cue, or vowel association. Consistent with Coppieters’ results, (i) the divergence of the advanced learners’judgments from NS norms was significantly greater than the variability exhibited by the Spanish NSs themselves, (ii) the learners’ judgements were less stable from speeded to unspeeded condition, and (iii) the sources of their judgements never completely replicated those of the NSs, even when their performance was native-like. Contrary to expectations, no differences were found between learners of higher and lower L2 proficiency, as measured by their scores on an elicited imitation (EI) measure (Bowden, 2007), or between their judgements and implicit or explicit language aptitudes, measured by LLAMA sub-tests D and F. In part 2, we briefly discuss seven problematic methodological issues surrounding research on the advanced learner: (1) conceptualizing and defining ‘advanced’; (2) measuring ‘advanced’ abilities using so-called “proficiency” tests and scales, EI or other procedures, impressionistically evaluated or scrutinized performance (Abrahamsson & Hyltenstam, 2009), or learners’ stage on a developmental scale (Bartning & Forsberg Lundell, 2018); (3) variability in native norms and in the non-native end-state; (4) mis-matches between L1 and L2 performance and underlying L1 and L2 competence; (5) the potential inadequacy of some individual difference measures for use with very advanced learners and/or with truncated samples (which makes predicting and explaining achievement of advanced abilities difficult); (6) choice of which linguistic items and types of items to target in achievement tests (e.g., optional or obligatory, from which linguistic and discourse domains?); (7) uncertainty over the comparability of L2 performance on various types of tasks under various conditions, e.g., grammaticality or acceptability judgment tasks, speeded or unspeeded, requiring implicit learning or implicit memory (Granena, 2018). We suggest potential temporary fixes for some of them, but note that all seven, and other issues, require definitive solutions.

To communicate effectively, it is crucial to be able to interpret the emotional state of one’s interlocutor (Ochs & Schieffelin, 1989), but it can be more challenging when the communication occurs in a second language or when not all communication channels are available (AUTHORS, 2018; Gendron et al., 2014). The present study investigates how first language (L1) speakers, second language (LX) speakers, and non-speakers (L0) of Mandarin perceive the emotional state of a Chinese interlocutor and whether communication modality affects this relationship. As most emotion research so far has focussed on Western languages (Barrett, 2011), the present study extends the area by investigating Westerners and non-Westerners’ perception of emotions expressed in Mandarin. It focusses on the perception of valence – i.e. how pleasant/unpleasant one is feeling – and arousal – i.e. how activated/calm one is feeling, which are assumed to be universal and semantically-primitive features of emotions (Russell, 2003). In total, 1599 participants (651 L1, 406 LX and 542 L0 Mandarin speakers) filled out an online survey embedded with 12 recordings of a Mandarin speaker acting out short emotional events. Participants were presented with each recording in one of the four investigated modalities, namely visual-vocal-verbal, vocal-verbal, visual-only, or vocal-only. After each recording, participants indicated their perception of how pleasant/unpleasant and how agitated/calm the actor was feeling via the Two-Dimensional Affect and Feeling Space (2DAFS) (AUTHOR in prep). The continuous valence and arousal ratings were analysed separately for each stimulus via bootstrapped regression models. For the valence ratings, a significant interaction between status of language (L1/LX/L0) and communication modality was revealed for eight out of 12 stimuli. Overall, communication modality did not affect valence ratings among L1 speakers – if we disregard the vocal-only modality which generally yielded neutral ratings, while it did affect the valence perception of LX and L0 speakers. For the arousal ratings, on the other hand, only two stimuli yielded a significant interaction between status of language and communication modality. Among the 10 other stimuli, communication modality does overall not affect arousal ratings regardless of the participants’ status of language. In sum, communication modality affects valence perception considerably more among LX and L0 speakers than among L1 speakers, while it does not greatly affect arousal ratings in general. These findings confirm that emotion perception, in particular valence perception, poses unique challenges for LX learners and that impeded modality can create communication challenges especially for LX users. Research in cross-linguistic emotion perception has valuable implications, among others for second language teaching. Focus on how to communicate emotions is absent from most classrooms while it is an integral part of communication (Dewaele, 2015). References Barrett, L. F. (2011). Constructing emotion. Psihologijske Teme, 20(3), 359–380. Dewaele, J.-M. (2015) On emotions in foreign language learning and use. The Language Teacher 39(3), 13-15. Gendron, M., Roberson, D., van der Vyver, J. M., & Barrett, L. F. (2014). Cultural relativity in perceiving emotion from vocalizations. Psychological Science, 25(4), 911–920. AUTHOR. (in prep). Two-Dimensional Affect and Feeling Space (2DAFS): a novel instrument to measure continuous affective dimensions and categorical feelings. AUTHORS. (2018). Emotion recognition ability across different modalities: The role of language status (L1/LX), proficiency and cultural background. Applied Linguistics Review. Ochs, E., & Schieffelin, B. (1989). Language has a heart. Text-Interdisciplinary Journal for the Study of Discourse & Communication Studies, 9(1), 7–26. Russell, J. A. (2003). Core affect and the psychological construction of emotion. Psychological Review, 110(1), 145–172. https://doi.org/10.1037/0033-295X.110.1.145