Session 1E

The last four decades of research on the bilingual lexicon point towards a unitary, nonselective lexicon. However, the role of some speaker- and word-level factors in shaping the lexicon and its functioning is less well known. The most prominent theoretical models differ in their conceptualization of bilingual semantic representations (localist vs distributed models). This project investigates bilingual lexical-semantic representation and processing employing cross-language visual priming. We focus on the role of word frequency and language use as continuous proxies of subjective frequency and explore potential modulations of bilingual lexical processing as a function of relative semantic overlap. Further, in a novel manner, we investigate the potential involvement of executive functioning in cross-language priming effects. 
            We tested 200 highly-proficiency Spanish-English bilinguals with varying degrees of L2 use in two cross-language priming lexical decision tasks with more than 400 word pairs. The large number of observations and a conservative analysis (α = .01 for main effects) with mixed-effects models (Baayen, 2008), ensured a robust dataset to explore several speaker- and word-level variables simultaneously. 
            We found significant effects of prime frequency and executive functioning in both translation directions. Significant effects of semantic alignment and language use were only present in the L1-L2 data. We discuss these results in light of theoretical models of the bilingual lexicon, and suggest new directions for future research.

Usage-based theories hold that language is dynamic, emergent, adaptive, variable and non-linear (Beckner et al. 2009), and that language development starts out from lexically specific utterances that gradually schematize (Tomasello 2003). In this paper, we combine methods from Dynamic Systems Theory (Verspoor, de Bot & Lowie 2011) and usage-based theory in order to investigate the development of epistemic constructions in L2 Norwegian. Epistemic constructions allow speakers to position themselves as knowledgeable, influence others' knowledge and mediate their social world, and can thus be viewed as vital for negotiating social identity (Ochs 1996) and demonstrating advanced L2 competence. The present study is part of a larger longitudinal investigation of adult learners of Norwegian on their way towards achieving B2-level certification. The data consist of dyadic, semi-structured conversations between two learners, Emilio (9 recordings) and Linda (8 recordings) and one of the researchers, collected over a period of 17 weeks. Norwegian epistemic expressions can be grouped into four categories: constructions with vet/tror 'know/think', modal verb constructions, constructions employing epistemic adjectives such as det er sant 'it/that is true', and epistemic adverbials. The four categories can be seen as functionally equivalent resources conveying similar meanings, which makes them eligible for studying as interacting components (cf. Verspoor & Van Dijk 2011). Subsequently, the data were subject to a type-token analysis, allowing for a detailed study of the learners' individual developmental trajectories. The present study reveals both similarities and differences in the individual developmental trajectories of Emilio and Linda. For both participants, vet/tror-constructions appear with high type/token-frequency already in the beginning of the study, indicating an ongoing schematization. At the same time, Emilio's vet/tror-constructions show growing variation, including tense inflection and word order, while Linda's vet/tror-constructions appear to be more fixed, with variation limited to the phrase complement. The data further suggest that while vet-constructions develop from a negated prototype jeg vet ikke 'I don't know', the contrary holds for tror-constructions. Constructions with epistemic adjectives appear late in the data, and seem to stand in a competitive relationship to epistemic adverbials. Adjectival constructions appear with low type frequency, but with growing token frequency of the specific exemplar det er sant 'that is true'. This finding suggests that adjectival constructions settle into an attractor state during the data collection period, functioning as "phrasal teddy bears" (Ellis 2012), particularly in Linda's L2. Modal verb constructions occur with marginal frequency, implying that such constructions come to use late in L2 development. Underscoring the value of longitudinal case studies of adult SLA, the two individual learning trajectories reveal similarities, possibly reflecting patterns in the input, but also differences: Construction development occurs unevenly, and differs in level of variability and non-linearity. Finally, we argue that the two learners are distinguished by both path and pace of schematization and their reliance on formulaic patterns. Beckner, C., Blythe, R., Bybee, J., Christiansen, M., Croft, W., Ellis, N. ... & Schoenemann, T. (2009). Language is a complex adaptive system: Position paper. Language learning, 59, 1-26. Ellis, N. (2012) Formulaic language and second language acquisition: Zipf and the phrasal teddy bear. Annual review of applied linguistics, 32, 17-44. Ochs, E. (1996) Linguistic resources for socializing humanity. In: Gumperz, J. & Levinson, S. (eds), Rethinking linguistic relativity. New York: Cambridge University Press, 407-437. Tomasello, M. (2003) Constructing a language: a usage-based theory of language acquisition. Cambridge, Mass: Harvard University Press. Verspoor, M., Bot, K. d. & Lowie, W. (eds) (2011) Dynamic Approach to Second Language Development: Methods and techniques, 29. Philadelphia: John Benjamins. Verspoor, M. & van Dijk, M. (2011) Visualizing interactions between variables. In: Verspoor, M., de Bot, K. & Lowie, W. (eds), 85-98.

This paper investigates the acquisition of the lack of actuality entailments of imperfective forms (Hacquard 2006, Homer 2011) by L2 learners of Spanish. While the preterit form triggers actuality entailments (i.e. the situation is understood as having taken place), imperfectives allow for both actuality entailment (e.g., progressive and habitual readings) and lack thereof. For example, the imperfective in (1) is compatible with Juan having sold the house (2) or not (3), as shown by its two congruent continuations. (1) Juan vendía-impf la casa. ‘Juan was selling the house’ (2) a. Juan vendía-impf la casa y la vendió. ‘Juan was selling the house and he sold it’ (3) a. Juan vendía-impf la casa pero cambió de opinión y no la vendió. ‘Juan was selling the house but he changed his mind and did not sell it’ In these examples the imperfect indicates that Juan had a plan to sell a house. Plans in the past can be argued to involve a more complex syntax (they include a modal phrase which we call PlanP) and more complex semantics (richer and more restricted-controllable etc.) We anticipate that this complexity, largely unnoticed in the literature, poses a challenge for English learners of Spanish because (i) the imperfective form is semantically vague, being compatible with two different scenarios, and (ii) classroom teaching focuses on the readings that trigger actuality (progressive and habitual) only. This implies that it may be difficult for English learners to be aware of this unusual meaning associated with the imperfect. To test this prediction, 60 L1 English learners of Spanish (20 beginners, 20 intermediate and 20 advanced) and 15 native speakers completed a context/sentence matching task and rated the appropriateness of two (imperfect/preterit) sentences in habitual, progressive and non-actual contexts (see examples above). Learners show high rates of correct acceptance of the imperfect and rejection of the preterit for the habitual interpretation (figure 1) and progressive (figure 2), showing that these imperfective readings are acquired adequately (i.e. the imperfect is not always persistently difficult for learners). Crucially, the results also show that the non-actual interpretation (the target context) exhibits very low rates of correct interpretation even at advanced levels. This finding is consistent with our hypothesis that the imperfective would be particularly difficult in this context. The differences in performance with respect of the different meanings of the imperfective suggest that the form itself is not the source of the difficulty for learners, it’s its association with multiple and varying semantic interpretations which poses an extra challenge (Montrul & Slabakova 2002; Domínguez et al. 2017). We argue that structures which need a syntax-semantics remapping in an L2 can be persistently difficult specially when the surface form in the L2 (e.g. the imperfect) does not match the surface form in the target language (e.g. the past progressive). Our results support the validity of the ‘Semantic Re-Distribution Hypothesis’ (Slababova & Montrul 2002; Montrul & Slabakova 2003; Domínguez, Arche & Myles 2017) which can easily account for why certain meanings associated with the Spanish imperfect are more difficult than others. References Domínguez, L., Arche, M. J. and Myles, F. (2017) Spanish Imperfect revisited: exploring L1 influence in the reassembly of imperfective features onto new L2 forms. SLR 33(4), 431-457. Hacquard, V. (2006). Aspects of Modality. PhD dissertation, MIT. Homer, V. (2011). Polarity and Modality. PhD dissertation, UCLA. Montrul, S., and Slabakova, R. (2002). The L2 acquisition of morphosyntactic and semantic properties of the aspectual tenses Preterite and imperfect. In The acquisition of Spanish morphosyntax (115-151). Netherlands: Kluwer. Slabakova, R., & Montrul, S. (2003). Genericity and aspect in L2 acquisition. Language Acquisition, 11(3), 165-196.

An increasing number of researchers have begun to use eye tracking during reading to explore how L2 learners (L2ers) apply grammatical knowledge during real-time processing, with a focus on how L2ers integrate lexical, morphosyntactic, semantic and discourse/pragmatic information to parse sentences (Felser and Cunnings, 2012; Clahsen et al., 2013; Jegerski and Sekerina, 2019). Relatively few studies, however, have investigated how secondary tasks influence processing patterns. A notable exception is Leeser, Brandl and Weissglass (2011) who found that certain secondary tasks (e.g. comprehension questions and judgment tasks) modulated L2ers’ processing of morphosyntactic violations during self-paced reading. The current study looks at how certain secondary tasks (e.g. comprehension questions vs. judgment tasks) influence L2ers’ processing of morphosyntactic violations, using data from two eye-tracking during reading experiments. To explore this, we focus on the processing of mood-modality mismatches in the L2 French of L1 English speakers. In French, certain verbs (such as vouloir ‘to want’) require the use of the subjunctive, as in (1) and (2). English, however, does not mark this distinction to the same extent. Previous research has shown that the French subjunctive is particularly challenging for L2ers (Ayoun, 2013; McManus and Mitchell, 2015). 1. Cécilia veut que son amie aitSUBJ plus de confiance en elle. Cécilia wants her friend to have more confidence in her. 2. * Cécilia veut que son amie aIND plus de confiance en elle. Cécilia wants her friend to have more confidence in her. Forty-five upper-intermediate-to-advanced English-speaking L2 French and 30 L1 French speakers completed a battery of tasks, including a background information questionnaire, independent proficiency measure (LexTALE-FR, Brysbaert, 2013) and two eye-tracking during reading experiments. In Experiment 1, participants’ eye movements were recorded as they read 36 test sentences and 36 filler items. The 36 items were split equally across two conditions: subjunctive and indicative (see sentences (1) and (2), respectively). After half of all trials, participants answered a comprehension question that focused on the content of the sentence. The stimuli in Experiment 2 was identical to Experiment 1. However, instead of comprehension questions, participants were asked to assess the acceptability of each sentence using a 7-point Likert scale (1=completely unacceptable, 7=completely acceptable). Mixed-effects linear regression models were computed for four eye-tracking measures (first fixation duration, gaze duration, go-past time and total reading time) on the critical region (i.e. the indicative/subjunctive-marked verb in the lower clause) in both experiments. These models revealed that L1 speakers consistently exhibited longer reading times for indicative sentences (versus subjunctive sentences) for each of the four eye-tracking measures, regardless of task type. L2ers, however, only exhibited significantly longer reading times for indicative sentences (compared with subjunctive sentences) for gaze duration (i.e. a measure sensitive to semantic and syntactic anomalies (Rayner et al., 2004) in the judgment task, as revealed by a significant interaction between condition and task (t = -2.58, p > 0.05). Proficiency did not appear to modulate this pattern. These findings reinforce Leeser et al.’s (2011) conclusions, suggesting that secondary tasks can influence L2ers’ processing of morphosyntactic violations not only in self-paced reading but also eye-tracking during reading. In the presentation, we discuss why this might be the case. More generally, our findings highlight the need for researchers to exercise caution when designing secondary tasks. In the absence of Experiment 2, we could have concluded that the L2ers’ processing was not target-like or that they had not acquired the property in question. It is clear that both tasks activate different knowledge sources, which ultimately has implications for researchers testing theories of L2 processing, such as the Shallow Structure Hypothesis.