Session 4A - Colloquium

One of the goals for second language (L2) learners is to attain automaticity, i.e., using a L2 quickly, efficiently and effortlessly. The questions of how L2 automatization can be most effectively fostered, however, have been under-researched in the field. We aim to lay out a research agenda for studying pedagogically useful questions by examining automatization in the context of L2 practice conceived of as deliberate and systematic activities for developing knowledge and skills (DeKeyser, 2007; Suzuki, Nakata & Dekeyser, 2019). 
This colloquium showcases five cutting-edge empirical studies on automatization that exemplify systematic investigation into "what practice activities are best for whom, for what structures, for which skill, at what time and in what context." (DeKeyser, 2018, p. xv) First, Conklin and colleagues examined how much and what type of modality (reading and reading while listening) led to more efficient processing of formulaic language, using an eye-tracking technique. Second, two studies focus on the issue of when we should schedule repeated practice to maximize automatization: Nakata and Elgort examined the effects of spaced practice for the acquisition of tacit (automatic retrieval of) vocabulary knowledge, while Hanzawa and Suzuki compared three schedules of task-repetition for oral fluency development in classroom. Last, two studies push the boundaries of research into the effects of form-focused instruction and the nature of information that can help automatization. In a semester-long classroom study, Oba examined what kind of pre-task practice (form-focused versus form-unfocused) should be integrated with communicative tasks for the automatization of morphosyntax by taking learners' language analytic ability and working memory capacity into account. McManus and Khoruzhaya investigated how much and what type of explicit information is needed for automatization of Spanish OVS-type sentences.
This colloquium hosts a variety of core research questions and covers diverse aspects of L2 practice and automatization: frequency and modality effects in incidental learning of formulaic language (binomials), deliberate learning for explicit and tacit vocabulary knowledge, distributed training for oral fluency development, the roles of systematic practice on morphosyntactic features and individual differences, and the effects of cross-linguistic information about word order on automatized sentence processing. Furthermore, the studies illustrate a variety of methodological options for measuring automatization (priming, lexical decision, picture interpretation, oral narration, and eye-tracking), using different research designs (short-term and longitudinal experiments in laboratory and classroom settings). This colloquium contributes to a deeper understanding of practice and automatization and its relevance to second language teaching and learning.

Differential effects of massed and distributed training on learning and retention have been observed in L1 and L2 learning and memory research. Studies in decontextualized vocabulary learning generally show that increased spacing between encounters with a given lexical item tends to facilitate its learning more than short or no spacing (e.g., Bahrick & Phelps, 1987; Karpicke & Bauernschmidt, 2011; Nakata, 2015; Nakata & Suzuki, 2019; Nakata & Webb, 2016), a phenomenon known as distributed practice effect. However, when unfamiliar vocabulary is encountered during reading, long spacing between repetitions may not necessarily facilitate learning because episodic memory traces of the previous contextual encounter may decay by the time the word is met again. Furthermore, whereas deliberate paired-associate studies usually involve learning and the rehearsal of the form-meaning associations, contextual learning is based on meaning inferences from context (i.e., inductive learning). Research on inductive learning (e.g., Carvalho & Goldstone, 2014; Kurtz & Hovland, 1956) also suggests that by juxtaposing multiple exemplars of a given word at once, massed learning (with no lag between encounters) may enable readers to discover the core meaning features of a new word more easily. The few contextual vocabulary learning studies that considered the effect of spacing have yielded inconsistent results regarding the benefits of spacing (Elgort, Brysbaert, Stevens, & Assche, 2018; Frishkoff, Collins-Thompson, Perfetti, & Callan, 2008; Serrano & Huang, 2018; Webb & Chang, 2015). The present study examined the effects of spacing on L2 vocabulary learning from sentence context. This study differs from earlier research in two major respects. Firstly, while previous contextual studies have compared the effects of short and long spacing, our study was designed to compare the effect of massed (no spacing) and spaced treatment. Second, to test the hypothesis that the effects of spacing differentially affect acquisition of explicit and tacit knowledge (that facilitates automatic lexical retrieval), spacing effects were measured using semantic priming and online lexical decisions, as well as explicit meaning recall and meaning-form matching posttests. Based on previous contextual word learning studies (Elgort et al., 2018; Elgort & Warren, 2014), as well as studies suggesting that long spacing is not necessarily effective for the knowledge proceduralization of grammar (Suzuki, 2017; Suzuki & DeKeyser, 2017) or pronunciation (Li & DeKeyser, 2019), we conjectured that the spacing effect may be present in the acquisition of explicit vocabulary knowledge but not necessarily in the acquisition of tacit knowledge of meaning. Sixty-six Japanese adult EFL learners read 144 English sentences with 48 novel vocabulary items. Each item appeared in three sentences and was assigned to either the massed or spaced condition. In the massed condition, the three sentences were presented simultaneously; in the spaced condition, one sentence was presented at a time, with 47 intervening sentences between the encounters (approximately 25 minutes). In both conditions, participants were asked to guess the meaning of the novel item, and were presented with the correct response as feedback. Mixed-effects modeling showed the advantage of spaced over massed learning on the meaning recall and multiple-choice posttest, and on the accuracy of online lexical decisions. No significant difference, however, was found between the massed and spaced distributions on semantic priming or degree of automatization of novel word processing. These results suggest that, although spacing facilitates acquisition of explicit vocabulary knowledge, massing may be as effective as spacing in the acquisition of tacit knowledge. We discuss the results of our study in terms of the learning and memory research and the mechanisms involved in contextual word learning, including contextual inferences, retrieval and feedback.

Recent L2 studies inspired by cognitive psychology research have investigated how manipulating the timing of repeated practice, without changing the total practice time, can enhance proceduralization of some aspects of L2 knowledge, such as grammar and pronunciation (e.g., Kasprowicz, Marsden, & Septhon, 2019; Li & Dekeyser, 2019; Suzuki, 2017). These findings point to the need to establish the optimal timing of task repetition for oral fluency development, which is highly dependent on proceduralization of linguistic formulation (Kormos, 2006). Recently, Bui, Ahmadian, and Hunter (2019) explored the effects of L2 task-repetition practice schedules on fluency gains. However, as the performance change was measured during the training phase only, the extent to which practice schedule influences fluency gain “retention” and “transfer” remains unknown. These shortcomings are overcome in the present study by investigating the effects of practice schedule on fluency development assessed by immediate and delayed transfer tests. For this purpose, a quasi-experimental design involving 80 English-as-a-foreign language learners attending four English classes at a Japanese university was adopted. Each of the classes was assigned to one of the four conditions (massed, short-spaced, long-spaced, and control). In the training session(s), the learners engaged in the narrative task consisting of the same six-frame cartoon six times, but with different speaking practice distribution. In the massed condition, the learners narrated the same cartoon six times consecutively (6 × 2 minutes), followed by a regular reading activity (45 minutes). Students assigned to the short-spaced condition narrated the cartoon three times in the beginning (3 × 2 minutes) and the end of a class (3 × 2 minutes), between which a regular 45-minute reading activity was inserted. Those in the long-spaced condition performed a set of three narrations (3 × 2 minutes) and the regular reading activity twice across one week. The learners in the control condition engaged in a regular reading activity only. Fluency gains were measured by a “transfer” narration test using novel cartoons, administrated immediately and one week after the training session(s). The pretest, immediate posttest, and delayed posttest yielded 240 speech datasets, which were analyzed in terms of speed fluency (mean length of run, articulation rate, and phonation-time ratio) and breakdown fluency (frequency and mean length of mid-clause and clause-final pauses). A series of repeated-measure ANCOVAs were conducted on each fluency measure with Condition (massed, short-spaced, long-spaced, and control) as between-participant factor, Time as within-participant factor, and pretest score as a covariate. Preliminary findings showed no significant main effect of Condition (p > .10) on any of the fluency measures; however, significant interactions between Condition and Time were detected on articulation rate and mid-clause pause frequency. Follow-up analyses suggested that, on the immediate posttest, learners assigned to the long-spaced condition had a significantly faster articulation rate compared to those in the massed and short-spaced conditions, whereas mid-clause pause significantly decreased in the massed condition compared to the other conditions. However, the main effects of Condition ceased to be significant on the delayed posttest. The present findings contribute to the current understanding of fluency training schedule effects (Bui et al., 2019) and suggest that manipulating the practice schedule can enhance fluency development on a novel transfer test. As the benefits were observed on the immediate posttest only, long-term effects on fluency development should be explored via a longitudinal intervention study. Key Reference: Bui, G., Ahmadian, M. J., & Hunter, A. M. (2019). Spacing effects on repeated L2 task performance. System, 81, 1?13. doi:10.1016/j.system.2018.12.006

Because of the inherent properties of language, it is theoretically possible to go a lifetime without producing or comprehending the same utterance twice (Pinker, 1999). Despite this possibility, we make use of a vast array of conventionalized word strings and sequences, commonly referred to as formulaic language. Conventionalized multi-word sequences account for between a third and a half of spontaneous discourse and in English may be as numerous as single words, possibly numbering into the hundreds of thousands (Erman & Warren, 2000). Because formulaic language is such an integral part of language, it is important to understand how it is acquired and how quickly a sensitivity to it becomes apparent in natural contexts in both a first (L1) and second (L2) language. Considerable research has demonstrated faster processing for formulaic sequences (Conklin, 2020), but little is known about the amount and type of exposure that is needed to develop this fast, efficient processing. The aim of this study was to explore the role of input modes, number of exposures, and textual enhancement on the processing of formulaic language. We focus on a particular type of formulaic language, binomials, which are sequences of x-and-y from the same part of speech. Their constituents are often semantic associates and are generally ‘fixed’, meaning that reversed form is much less frequent (i.e. it is generally ‘salt and pepper’ not ‘pepper and salt’). While they can be literal or figurative, here we focus on fully compositional, literal binomials. In the study, 50 L1 and 50 L2 participants encountered two types of binomials, existing (‘time and money’) and novel (‘wires and pipes’) ones, in short stories of approximately 1,500 words that were presented in a reading-only (RO) mode or a reading-while-listening- (RWL) mode. The existing binomials occurred two times while the novel ones occurred two, four, five or six times. The binomials were either unenhanced or enhanced (appeared in red). In the enhancement condition, all occurrences of the binomial appeared in red. To monitor the development of a sensitivity to novel binomials, L1 and L2 readers’ eye movements were monitored while they read stories containing them. Both L1 and L2 readers spent more time reading the binomials (longer and more fixations) in RWL than RO and this was more apparent in the enhanced condition. Enhancement more readily captured the attention of the L2 readers, with L2 speakers having more and longer fixations to enhanced binomials than the L1 speakers. An important comparison was reading behavior for existing and novel binomials; as novel binomials become entrenched in memory, a sensitivity to them should arise automatically and their reading pattern should become similar to that of existing binomials. At the second occurrence existing binomials were processed more quickly (shorter and fewer fixations) than novel ones. In the enhanced condition, processing was the same for existing and novel binomials for L1 readers after four occurrences and by six for L2 readers. These results show that in as little as four exposures for L1 speakers and six for L2 speakers, novel binomials are processed as fast as existing ones. Crucially, the findings demonstrate the importance of multiple exposures as well as the benefit of input enhancement for the development of a sensitivity to lexical patterns (i.e. formulaic language), which arises rapidly and automatically in the context of natural reading. References Conklin, K. (2020). Processing single- and multi-word items. In S. Webb (Ed). Routledge Handbook of Vocabulary Studies. Routledge. Erman, B. and Warren, B. 2000. ‘The idiom principle and the open-choice principle,’ Text 20/1: 29-62. Pinker, S. 1999. Words and Rules: The Ingredients of Language. New York: Harper Collins.