Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice [1st ed.] 978-3-030-22065-5;978-3-030-22066-2

Table of contents :
Front Matter ....Pages i-xxiv
Cross-Linguistic Influence in Second Language Speech: Implications for Learning and Teaching (Jaydene Elvin, Paola Escudero)....Pages 1-20
/ðə ˈmusɪk ɪnˈdustrɪ jas esˈtarted teikin leˈgal akˈʃɒn/*. A Preliminary Study on the Nature and Impact of Phonological and Orthographic Transfer in the English Speech of Bilingual Speakers of Spanish and Galician (Yolanda Joy Calvo-Benzies)....Pages 21-47
CLI in Lexical Accessibility (María del Pilar Agustín-Llach)....Pages 49-63
Cross-Linguistic Influence at the Level of Word Order in L3 English by L1 Georgian/L2 Russian Speakers (Gvantsa Jichoshvili, M. Juncal Gutierrez-Mangado)....Pages 65-86
Didactic Challenges in the Multilingual Classroom—The Case of French as a Foreign Language (Christina Lindqvist)....Pages 87-99
Syntactic Transfer in L3 Learning. What Do Models and Results Tell Us About Learning and Teaching a Third Language? (Camilla Bardel)....Pages 101-120
What Lies Beneath: L1 Morphosyntax Seeping in Through Young Learners’ EFL (Raúl Azpilicueta-Martínez)....Pages 121-148
“I’m Jealous but I Am Very Happy”: Congratulating in an EFL Context (M. Luz Celaya, Laura Panelli, Júlia Barón)....Pages 149-167
The Use of Communication Strategies in L3 English CLIL Learners (Izaskun Arratibel-Irazusta, María Martínez-Adrián)....Pages 169-189
Explicit Plurilingualism in Co-taught CLIL Instruction: Rethinking L1 Use (Iris Milán-Maillo, Elisabet Pladevall-Ballester)....Pages 191-213
Native Language Identification by Human Judges (Scott Jarvis, Rosa Alonso Alonso, Scott Crossley)....Pages 215-231
Back Matter ....Pages 233-265

Citation preview

Second Language Learning and Teaching

M. Juncal Gutierrez-Mangado María Martínez-Adrián Francisco Gallardo-del-Puerto Editors

Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice

Second Language Learning and Teaching Series Editor Mirosław Pawlak, Faculty of Pedagogy and Fine Arts, Adam Mickiewicz University, Kalisz, Poland

The series brings together volumes dealing with different aspects of learning and teaching second and foreign languages. The titles included are both monographs and edited collections focusing on a variety of topics ranging from the processes underlying second language acquisition, through various aspects of language learning in instructed and non-instructed settings, to different facets of the teaching process, including syllabus choice, materials design, classroom practices and evaluation. The publications reflect state-of-the-art developments in those areas, they adopt a wide range of theoretical perspectives and follow diverse research paradigms. The intended audience are all those who are interested in naturalistic and classroom second language acquisition, including researchers, methodologists, curriculum and materials designers, teachers and undergraduate and graduate students undertaking empirical investigations of how second languages are learnt and taught.

More information about this series at http://www.springer.com/series/10129

M. Juncal Gutierrez-Mangado María Martínez-Adrián Francisco Gallardo-del-Puerto

•

•

Editors

Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice

123

Editors M. Juncal Gutierrez-Mangado Department of English and German Philology and Translation and Interpreting, Faculty of Arts University of the Basque Country UPV/EHU Vitoria-Gasteiz, Spain

María Martínez-Adrián Department of English and German Philology and Translation and Interpreting, Faculty of Arts University of the Basque Country UPV/EHU Vitoria-Gasteiz, Spain

Francisco Gallardo-del-Puerto Department of Philology University of Cantabria Santander, Cantabria, Spain

ISSN 2193-7648 ISSN 2193-7656 (electronic) Second Language Learning and Teaching ISBN 978-3-030-22065-5 ISBN 978-3-030-22066-2 (eBook) https://doi.org/10.1007/978-3-030-22066-2 © Springer Nature Switzerland AG 2019 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

To Balentin, for the pleasure of flying by his side To Adriana, my inspiring diamond To Inés, for there’s so much of her in me

Foreword

A major challenge when investigating the acquisition of a new target language (TL)1 concerns the determination of the role of other languages known in the developmental process. Attempts to quantify this so have a long history. One early approach, contrastive analysis (CA) (Fries, 1945; Lado, 1957) claimed that the learning of a new TL, like the first language (L1) acquisition, consisted of the learning of a fixed set of language “habits” over time. In TL acquisition, it was argued that the learner attempts to “transfer” the linguistic habits from the L1 to the TL. Where the L1 and TL match, the positive transfer takes place and facilitation in learning occurs. Where the linguistic habits between the L1 and the TL do not match, there is a negative transfer of habits. This results in the disruption in learning and interference (see extended discussions in Flynn, 1987; Flynn & O’Neil, 1988; Epstein, Flynn, & Martohardjono, 1996). In order to remedy the disruption, the learner must acquire the new habits of the TL through modification of the L1 habits, for example, by addition or deletion (see early discussions in Gass (1988), Newmeyer and Weinberger (1988) and Rutherford (1988)). This approach was articulated within a behaviorist framework for language learning. As noted by Flynn and O’Neil (1988, p. 6), “stripping away (this approach) from its behaviorist foundations, we can see that it captured an important aspect of the L2 [TL] acquisition process—the role of the L1 knowledge in L2 [TL] acquisition”. Yet, as we all know, this approach failed to provide an explanatory theory of the TL acquisition process. It failed to reliably predict when “interference” would occur and when “facilitation” would occur although it did have “some” relative success when dealing with cross-linguistic influence (CLI) phonologically. Another major problem concerns the use of the metaphor of “transfer”. Even though Fries and Lado did attempt to empirically quantify this notion, it essentially remains as a very weak and unfortunate metaphor used to describe the processes underlying the acquisition of a new TL. What does it mean to essentially delete the habits of the L1 1

This target language may be the second (L2), the third (L3), the fourth (L4), etc. The papers in this chapter, as will be noted, focus primarily on L2 and L3 acquisition by monolingual and bilingual speakers.

vii

viii

Foreword

when there is negative transfer from the L1 to the TL? Does the L1 grammar lose these language structures in that it then becomes less of a language itself? How exactly is the TL grammar constructed? Is it constructed piecemeal with some L1 habits and/or with modified L1 habits? What does such a TL grammar generate? We all know that “transfer” cannot be taken literally as an essential process in the construction of a new TL grammar; yet, such terminology persists. CLI as used in this book is a far more acceptable and descriptively accurate manner at this point in time to capture what was traditionally known as “transfer”. It is important to note once again that while CA failed to provide a wholly explanatory model of TL acquisition, it did capture some aspect of the TL learning process but what this is precisely continues to elude the field. We know, for example, that the acquisition of English as a TL by a Japanese speaker and a Spanish speaker, all else being equal, is not identical at all levels. The Spanish speaker has a developmental advantage over the Japanese speaker in terms of the “match” of certain deep abstract properties of the grammars of English and Spanish, and most notably they match in head direction (head-initial) structure, while that of Japanese (head-final) does not (see, e.g., Flynn, 1987, 2018). This match benefits the Spanish speaker and allows the speaker to not to have to redundantly represent a head-initial complement structure in the mind/brain. It already exists. However, the Japanese speaker learning English must represent the head-initial complement structure for English for the first time in much the same way that a young child learning English as an L1 must do. An alternative model offered in 1974 concerned the development of the creative construction (CC) theory of L2 acquisition (Dulay & Burt, 1974). This approach, as documented widely in many other sources, attempted to capture the fact that much of the L2 or TL acquisition process shared certain fundamental properties noted in the L1 acquisition process. That is, many of the developmental patterns isolated in L1 acquisition were consistent with those noted for TL acquisition. For early work in this regard, see Bailey, Madden and Krashen (1974), Flynn (1987, 1988), Mazurkewich (1988), Epstein, Flynn and Martohardjono (1996) among many others. Such findings led researchers at the time to hypothesize that both L1 and TL acquisition follow from a set of common principles unique to the language learning process. It was argued that the principles were, by and large, cognitively unique— i.e., linguistic in nature—in that both children and adults evidence them. Children are still developing cognitively in many ways, and adults have supposedly reached a level of steady-state cognition. Thus, the common, unifying factor for both children and adults is the developing language grammars. Both children and adults uncannily resembled each other in acquisition. This development cannot be inextricably tied to the development of general cognitive processes, as we would not expect then to see the similarities across the two populations as the adults have reached a steady state in other domains of cognition. The CC argued that the role of other languages known would emerge in only very superficial ways. What this actually meant was never entirely developed. While it did capture an aspect of the TL acquisition process, namely that L1 and TL acquisition derive from a common

Foreword

ix

set of principles in many ways and at many levels, it still failed to account in an explanatory manner for any form of CLI when it seemed to occur. What emerges from all this early research is that there was a need to understand in a principled and explanatory manner ways in which there might be some role for other languages known and at the same time ways in which both L1 and L2 are fundamentally the same. Universal grammar–generative linguistics was proposed as a possible solution or way to unify both the CA and the CC properties of the language learning process (see Flynn, 1987, 2018). Although my purpose here is not to develop this theory, I simply note that the debates continue to this day in terms of attempting to quantify in a principled manner ways in which other languages known emerge in the language acquisition process and this is where this book enters into the debates. This book entitled Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice attempts to deal with the issues of CLI in very unique and promising ways. The papers in this volume seek to determine the role of “cross-linguistic influence” in both the classroom and the naturalistic settings in distinct domains. The areas of study covered in this book are broad and are ones that are not always reported on together in one volume: phonology (Elvin & Escuerdo; Calvo Benzies), lexical access (Agustín Llach), pragmatic influences (Celaya, Panelli & Barón), syntax (Bardel; Azpilicueta Martinez; Jichoshvili & Gutierrez-Mangado), classroom learning (Maillo & Pladevall Ballester; Lindquist), child TL acquisition (Azpillicueta Martinez; Millán Maillo & Pladevall Ballester), adult TL (Escudero & Elvin; Calvo Benzies; Agustín Llach; Jichoshvili & Gutiérrez-Mangado; Bardel; Celaya, Panelli & Barón; Arratibel-Irazusta & Martínez-Adrián; Jarvis, Alonso & Crossley), L3 learning (Bardel) to name a few. The range of languages covered is also wide ranging: Catalan, Spanish, Swedish, Finnish, Russian, Galician, English, etc. In addition, the range of proficiency levels investigated is broad. Similarly, there are experimental and non-experimental papers with the majority being empirical studies. And, the empirical papers are based on distinct experimental tasks: grammaticality judgment task (GJT), rater-based judgments of written text, etc. Importantly, however, in addition to the discussion of CLI in TL acquisition, each chapter discusses the results presented in terms of their pedagogical implications. While the papers are not generally articulated within a particular framework with the exception of Escudero and Elvin’s and Bardels’s, they provide important new empirical evidence that ultimately must be incorporated into some theory or explanatory framework. One of the most striking characteristics of this edited volume is the fact that overall, the empirical results reported are mixed both across the studies reported and even within a particular study. For example, Escudero and Elvin report CLI in TL acquisition, at least at initial stages, and Llach does not. And within the word order results reported here, both CLI and non-CLI in the acquisition of new TLs are

x

Foreword

indicated. Word order acquisition did not pose a problem for the bilinguals suggesting a bilingual advantage as hypothesized by the authors. More specifically, this book consists of 11 chapters, each dealing with a question or hypothesis concerning some aspect of CLI in either the classroom or in an experimental context. To begin, in Chapter “Cross-Linguistic Influence in Second Language Speech: Implications for Learning and Teaching”, Escudero and Elvin seek to determine the extent to which the Second Language Linguistic Perception (L2LP) model can account for the perception, comprehension and production of the sounds of a new target language. The authors argue that because “phonemic inventories differ across languages and dialects, cross-linguistic difficulty is not uniform”. Depending upon the match/mismatch of the phonemic inventories of the languages or dialects known by a learner, some aspects of the TL may be more difficult for some learners and not others depending on the L1. The central claim is that “learners will initially perceive and produce the sounds of the L2 and recognize L2 words in the same way that they do in their own native language”. The authors predict that problems in TL acquisition will be learner-specific dependent on the “phonetic properties of the native and target language dialects”. They present a review of studies to support their model. The review is extensive although limited to those that support the model. What is important with this study is the claim that the results will be learner-specific depending upon the prior linguistic experience and knowledge of the learner. Each learner could potentially present with a unique constellation of prior knowledge that must be taken into consideration. I do not believe the authors would argue that there is no systematicity in patterns of acquisition. The patterns must be understood and considered in light of the match/mismatch of the properties of the specific languages that the learner presents with. In Chapter “/ðə ˈmusɪk ɪnˈdustrɪ jas esˈtarted teikin leˈgal akˈʃɒn/*. A Preliminary Study on the Nature and Impact of Phonological and Orthographic Transfer in the English Speech of Bilingual Speakers of Spanish and Galician”, Calvo Benzies presents a preliminary study on the nature and impact of phonological and orthographic transfer in the language of bilingual speakers of Spanish and Galician. The paper aims to describe instances of phonological and orthographic CLI in the spoken language of Galician and Spanish speakers in their acquisition of English. Three spoken language tests were employed; both high school- and university-level participants were tested. The author notes significant CLI in either phonological or orthographical “interference” from the speakers’ native languages. However, not every phonological or orthographic contrast between English and the participants’ native languages presented problems. The “interferences” were very focused and specific. This is one study in which the results were mixed; i.e., the author reports both CLI and non-CLI in acquisition. In Chapter “CLI in Lexical Accessibility”, Agustín Llach seeks to evaluate “the role of L1 [CLI] when completing a lexical availability task”. The participants were Spanish pre-university students learning English as a new TL. This paper raises important issues concerning what is available to the L2 learner and what is used in

Foreword

xi

the essential construction of a new TL. Negative CLI is very rare as reported. The authors propose models for which their data are compatible. Important discussion ensues concerning implications for the classroom. In Chapter “Cross-Linguistic Influence at the Level of Word Order in L3 English by L1 Georgian/L2 Russian Speakers”, Jichoshvili and Gutierrez-Mangado investigate word order concerning the OV/VO alternation in main and embedded clauses by Georgian (OV word order) and Russian (VO word order) speakers learning L3 English (VO word order). Three groups of learners were tested: Georgian children and adults and L1 Georgian/L2 Russian bilingual children. All participants were at an A2 proficiency level. The learners were administered a grammaticality test and a translation task involving affirmative and negative stimuli. Results indicate no statistically significant differences among the groups on either task. The authors argue that there were no signs of transfer at the A2 level. Monolingual and bilingual speakers have acquired the word order pattern of the L2/L3; this was not a source of difficulty for the learners. However, there seemed to be a bilingual advantage in that the L1 Georgian speakers were not as good at rejecting ungrammatical sentences as accepting grammatical sentences in the GJTs. The L1 Georgian/L2 Russian were equally accurate on both types of sentences suggesting some form of a bilingual advantage given their knowledge of two different languages. If so, this result is consistent with those results reported by Berkés and Flynn (2017) and Flynn (2018) in which knowledge of multiple languages known can positively influence the subsequent language development. In Chapter “Didactic Challenges in the Multilingual Classroom—The Case of French as a Foreign Language”, Lindqvist investigates the multilingualism of Swedish learners of French (an L3 for most of the participants aged 12–16 years old). Many of these participants in the study may also have studied other foreign languages as well, viz., Spanish or German. In written retellings of a story by the Swedish pupils, the results indicated that some of the other languages known or studied formally by the students were used when the participants wrote in French. Lindqvist argues that these students may be undergoing “learning processes in several languages simultaneously”. The author raises important questions and highlights the importance of these results in terms of challenges faced by educators in the modern-day multilingual classroom. In Chapter “Syntactic Transfer in L3 Learning. What Do Models and Results Tell Us About Learning and Teaching a Third Language?”, Bardel notes that investigation into L3 acquisition is a relatively new area of study that has grown in interest and importance. Results of L3 investigations can provide insights about the acquisition process that the study of L1 or L2 acquisition alone cannot provide (see discussion in Flynn, Foley, & Vinnitskaya, 2004 as well as, e.g., 2017). The review of the models presented here indicates, as argued elsewhere, that both the L1 and the L2 can play a role in subsequent L3 acquisition. However, the precise nature of which structures play a role in CLI in L3 learning “remains unsolved”. Critically, much more research is needed in terms of replication and confluence of results across both comparable and distinct future studies. This chapter concludes with an

xii

Foreword

important discussion of the possible pedagogical implications of each of these models. In Chapter “What Lies Beneath: L1 Morphosyntax Seeping in Through Young Learners’ EFL”, Azpilicueta Martínez addresses the question concerning the notion of negative correlation between proficiency and lexical CLI. Twenty L1 Spanish speaking children aged 8;00 learning English were investigated in a content and language integrated learning (CLIL) program. The learners carried out a task first with an expert speaker of English and subsequently with a level-matched peer. Overall, results revealed “an extremely low rate of explicit L1 use, yet revealed the existence of more pervasive CLI in the form of structural transfer, particularly so when children interacted with their peers”. In Chapter ““I’m Jealous but I Am Very Happy”: Congratulating in an EFL Context”, Celaya, Panelli and Barón tackle an issue that has received little empirical attention in traditional CLI studies: speech acts concerning congratulations. The study seeks to analyze the effects of age, and proficiency and “the influence of pragmatics of the L1 in the production of this speech act (congratulations) by learners of English as a foreign language in a school context with no instruction on pragmatics”. Participants were administered a written discourse completion task consisting of four different situations in which the participants had to congratulate the interlocutor. Results indicate that young and less proficient learners tend to use “units around the actual congratulation chunk that involve a focus on themselves rather than on the interlocutor”. Older more proficient learners used categories “such as offers of good wishes and hopes” as well as statements suggesting actions and encouragements. The authors argue that these expressions directly involve the interlocutor as would be expected in this speech act. The differences between the two groups are discussed in terms of age and proficiency level. Results suggest that more systematic study is needed to understand CLI in the acquisition of language-specific pragmatics. In Chapter “The Use of Communication Strategies in L3 English CLIL Learners”, Arratibel-Irazusta and Martínez-Adrián focused on “the use of communication strategies in an oral narration task and its correlation with general proficiency and knowledge of receptive vocabulary in two different age/proficiency CLIL groups of the third language (L3) English learners”. Results indicate that proficiency level does not play a major role in the use of communicative strategies as very few differences are found among the groups analyzed. Correlational results indicate that although previously known “language-based strategies do seem to go hand in hand with the level of proficiency and receptive vocabulary, this does not seem to apply to the use of TL-based strategies”. It is also not the case that the use of TL-based strategies implies a “lower use of prior linguistic experience”. The use of previously known languages is still important as the L3 learners “seem to use it to scaffold L3 production”. In Chapter “Explicit Plurilingualism in Co-taught CLIL Instruction: Rethinking L1 Use”, Millán-Maillo sought to answer the question about whether “use of the L1 in L2 settings can facilitate L2 learning” as is often claimed. Catalan/Spanish bilingual primary school learners were investigated in English co-taught CLIL

Foreword

xiii

science sessions. The main findings indicated that “this specific co-teaching model did not fulfill the objectives of CLIL approaches but L1 use was shown to be a beneficial tool serving the purpose of coping with CLIL linguistic and cognitive demands”. In Chapter “Native Language Identification by Human Judges”, the authors, Jarvis, Alonso and Crossley, report the results of “two studies involving native language identification by human judges”. The first study involved six Finnish speakers who were given 16 texts written in L2 English. Half the texts were written by Finnish speakers and half by Swedish speakers. The raters were asked to read each text and make a judgment about whether it was written by a Finnish or a Swedish speaker. The second study includes the same six Finnish speakers and an additional group of ten Spanish speaker raters. Results indicate that some of the raters were accurate above 80% for top Finnish and 90% for top Spanish. Implications of these results are discussed in terms of their meanings for the classroom. In addition to a discussion concerning the pedagogical implications of these results, the authors also focus on isolating those features of the texts that resulted in such high-accuracy scores. In conclusion, this book is a very important one as it provides a fresh look at CLI in new contexts. It contributes to our growing knowledge concerning the role of CLI in the TL development. It replicates at a very general level the fact that in some cases CLI emerges and in others it does not. This book provides multiple frameworks for continued empirical testing. It offers a “breath of fresh air” to classical CLI investigations, and it is innovative in the range of papers chosen for publication. At another level, the results reported in this book highlight the fact that we are still a long way from understanding the human cognition for language. This is a very challenging task to take on as it is well known that in behavioral investigations of the language development, we must always make inferences from what we observe with the performance data collected—which itself is fraught with problems in terms of understanding the underlying developing competence for language. Subsequent studies must consider and acknowledge the fact, as isolated here with these studies, that CLI is not a uniform phenomenon that emerges uniformly in all linguistic domains and under all conditions. There are multiple factors that influence the outcome of the studies. Some of these factors concern the match/mismatch of the languages known by the learners as suggested by studies in this book and elsewhere, the domain of study investigated, and all aspects of the experimental design, viz., the proficiency levels of the participants and the nature of the investigative tool, the stimulus items and their controls. Moving forward, careful attention must be paid to replication of the studies in CLI and, importantly, not just for CLI studies but for all acquisition studies as is done in all science. Finally, this book offers a unique set of studies all focused on determining the role of other languages known in the acquisition of subsequent new target languages. The breadth of languages and studies is new and interesting. This book provides a mixture of experimental studies and classroom-based studies that are important and allow immediate comparisons across the different forms of investigation. Results provide new ways to look at CLI; they also raise new questions.

xiv

Foreword

Clearly, the results do not satisfy the requirements of a strict CA approach to TL learning nor do they completely satisfy a CC model. The fundamental question remains, “How does the learner construct the grammar of the new TL(s)”? Understanding this has consequences for theories of cognition generally and language acquisition specifically. Knowledge of how this is accomplished allows us to know what has to be taught to our students and what does not. It essentially allows us to know what we can assume about the language knowledge bases of our language learners and what we cannot. Many questions remain, but it is important to once again state that this innovative set of papers makes an important and unique contribution to the field of TL learning, especially with respect to CLI in this process. Cambridge, MA, USA

Suzanne Flynn MIT Linguistics and Philosophy [email protected]

Acknowledgements

We are grateful to all the authors who have generously contributed to this volume: without their commitment, this book would not have seen the light. Special thanks to Suzzane Flyn who kindly agreed to write the foreword. We also want to thank all the reviewers whose insightful comments and suggestions have contributed to this book: Tania Angelovska, Anna Balas, Peter Ecke, Ana Fernández Dobao, Evelyn Gandón Chapela, María del Pilar García Mayo, Britta Hufesien, Carol Jaensch, Amparo Lázaro Ibarrola, Angel Lin, Cristobal Lozano, Natalia Shervin Malmasi, Natalia Mitrofanova, Jean Carles Mora, Mireia Ortega Durán, Simone Pfenninger, Jorge Pinto, Eloi Puig-Mayenco, Lucrecia Rallo Fabra, Pilar Safont, Gila Schauer, Marit Westtergaard, John Witney and Magdalena Wrembel. We are also grateful to the Springer Second Language Learning and Teaching Series Editor Miroslaw Pawlak, Ravi Bengadachalam and Arulmurugan Venkatasalam for their support and help in the process of editing this book. We would also want to thank the financial support provided by the Department of Education, Universities and Research of the Basque Government (IT904-16), University of the Basque Country UPV/EHU (Grant UFI 11/06) and the Spanish Ministry of Economy and Competitiveness (MINECO FFI2016-74950-P; MINECO FFI2012-32212 and MINECO FFI2012-34214).

xv

Introduction

Bi-/multilingual speakers are by definition those speakers who have knowledge of more than one language. Accommodating more than one language in the brain means that at any time the bi-/multilingual speaker can potentially access these languages whenever he/she needs to make use of one of them in order to understand or convey a message. Research has shown that the languages in the bi-/multilingual individual are not isolated from each other and consequently may interact in a myriad of ways as has been observed in speakers who are highly competent in all the languages they know (e.g., in cases of code-switching). Interaction between languages can also take place when a speaker is in the process of acquiring an additional language, that is, when the knowledge of the additional language is not complete. This type of interaction is known as cross-linguistic influence (CLI) and is the topic of this book. This term has been widely used as a theory-neutral term to encompass notions such as “transfer”, “interference”, “avoidance”, “borrowing” and aspects of language loss (Sharwood-Smith & Kellerman, 1989, p. 1).2 CLI has been viewed as either a learning or a communication strategy and can potentially have a big impact on any aspect of language, from phonetics/phonology to morphosyntax, semantics and pragmatics. CLI can be constrained by a series of factors ranging from proficiency in the L1 and or L2/Ln, similarities/differences between the L1 and the L2/Ln, age of acquisition of the L1/Ln, recency, among others. Interesting and necessary as research on CLI is, the knowledge obtained about CLI needs to be ultimately accessible to language teaching practice, and this is an area where advances have been profusely lacking. The reasons for this unsatisfactory situation are numerous, as the authors in the volume make clear: Research findings are mostly targeted toward specialized audiences, rarely toward language teachers, which makes the formal language used difficult to access for the latter. The lack of trained teachers to teach pronunciation and the lack of classroom-oriented research (most research is laboratory-based) also make the findings obtained not fully generalizable to the rest of the population. 2

But see Gonzalez Alonso and Rothman (2017) for differences between notions such as transfer and CLI.

xvii

xviii

Introduction

In any case, L2 teaching and learning practices which take into account how CLI in particular language combinations can influence the process of acquisition will undoubtedly prove to be more effective than when this knowledge is lacking. Practitioners who have knowledge of CLI will be able to gear their teaching practice more effectively toward learners whose language combinations are known, as every chapter in the volume makes explicit in different linguistic domains. The need to reduce the differences that separate theory and practice has been voiced by numerous researchers in the past but has met with limited responses, and this book is an attempt to remedy this situation. The idea for a book on CLI has long been in our minds, but it was only after the success of the workshop organized by the LASLAB research group (http://wwwl. laslab.org) in Vitoria-Gasteiz on April 27, 2016, that the idea began to take real shape. In addition to the pedagogical implications offered in light of the empirical findings obtained in each study included in the book, another goal has been to incorporate studies depicting a wider array of contexts and scenarios (i.e., L2 and L3 learning situations, CLIL), a wider range of topics and perspectives (i.e., perspective of the speaker and the listener, directionality effects between unrelated languages, task effects), as well as language levels, so as to offer a better picture of the complexity of multilingualism, which is more the norm than the exception nowadays. In addition, taking into account the need of research with children and the increasing popularity of early foreign language teaching programs (García Mayo, 2018), we have included studies targeting child as well as adult learners. This book is divided into eleven chapters, and each chapter finishes with a section where the authors make explicit reflections on how the results presented/discussed can be of use to teachers in the classroom. This book is intended for SLA researchers, graduate students in SLA research and applied linguistics (MA, Ph.D.) as well as language teachers, as the various chapters provide the state-of-the-art research results on different areas of CLI together with explicit and research-informed pedagogical implications which can be taken up by educators to make the most of CLI. We hope that the chapters presented will call the attention of language practitioners and that the pedagogical implications depicted in the different chapters see the light in the daily classroom life in the near future. Vitoria-Gasteiz, Spain Vitoria-Gasteiz, Spain Santander, Spain

M. Juncal Gutierrez-Mangado María Martínez-Adrián Francisco Gallardo-del-Puerto

Contents

Cross-Linguistic Influence in Second Language Speech: Implications for Learning and Teaching . . . . . . . . . . . . . . . . . . . . . . . . Jaydene Elvin and Paola Escudero /ðə ˈmusɪk ɪnˈdustrɪ jas esˈtarted teikin leˈgal akˈʃɒn/*. A Preliminary Study on the Nature and Impact of Phonological and Orthographic Transfer in the English Speech of Bilingual Speakers of Spanish and Galician . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yolanda Joy Calvo-Benzies CLI in Lexical Accessibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . María del Pilar Agustín-Llach

1

21 49

Cross-Linguistic Influence at the Level of Word Order in L3 English by L1 Georgian/L2 Russian Speakers . . . . . . . . . . . . . . . . . . . . . . . . . . Gvantsa Jichoshvili and M. Juncal Gutierrez-Mangado

65

Didactic Challenges in the Multilingual Classroom—The Case of French as a Foreign Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Christina Lindqvist

87

Syntactic Transfer in L3 Learning. What Do Models and Results Tell Us About Learning and Teaching a Third Language? . . . . . . . . . . . . . . 101 Camilla Bardel What Lies Beneath: L1 Morphosyntax Seeping in Through Young Learners’ EFL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 Raúl Azpilicueta-Martínez “I’m Jealous but I Am Very Happy”: Congratulating in an EFL Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 M. Luz Celaya, Laura Panelli and Júlia Barón

xix

xx

Contents

The Use of Communication Strategies in L3 English CLIL Learners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 Izaskun Arratibel-Irazusta and María Martínez-Adrián Explicit Plurilingualism in Co-taught CLIL Instruction: Rethinking L1 Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Iris Milán-Maillo and Elisabet Pladevall-Ballester Native Language Identification by Human Judges . . . . . . . . . . . . . . . . . 215 Scott Jarvis, Rosa Alonso Alonso and Scott Crossley References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261

Contributors

María del Pilar Agustín-Llach Universidad de La Rioja, Logroño, La Rioja, Spain Rosa Alonso Alonso Facultad de Filología y Traducción, Departamento de Filología Inglesa, Francesa y Alemana, Universidad de Vigo, Vigo, Spain Izaskun Arratibel-Irazusta Department of English and German Philology and Translation and Interpreting, Faculty of Arts, University of the Basque Country UPV/EHU, Vitoria-Gasteiz, Spain Raúl Azpilicueta-Martínez Departamento de Ciencias Humanas y de la Educación, Universidad Pública de Navarra (UPNA), Pamplona, Spain Camilla Bardel Institutionen Stockholm, Sweden

för

Språkdidaktik,

Stockholms

Universitet,

Júlia Barón Institute for Multilingualism at the Universitat Internacional de Catalunya, University of Barcelona, Barcelona, Spain Yolanda Joy Calvo-Benzies Departamento de Filología Española, Moderna y Clásica, University of the Balearic Islands, Palma de Mallorca, Spain Scott Crossley Department of Applied Linguistics/ESL, Georgia State University, Atlanta, GA, USA Jaydene Elvin Department of Linguistics, California State University, Fresno, CA, USA Paola Escudero The MARCS Institute for Brain, Behaviour and Development and the Australian Research Council Centre of Excellence for Dynamics of Language, Western Sydney University, Penrith, NSW, Australia M. Juncal Gutierrez-Mangado Department of English and German Philology and Translation and Interpreting, Faculty of Arts, University of the Basque Country UPV/EHU, Vitoria-Gasteiz, Spain

xxi

xxii

Contributors

Scott Jarvis Department of Linguistics, University of Utah, Salt Lake City, Utah, USA Gvantsa Jichoshvili University of the Basque Country UPV/EHU, VitoriaGasteiz, Spain Christina Lindqvist Department of Languages and Literatures, University of Gothenburg, Gothenburg, Sweden M. Luz Celaya University of Barcelona, Barcelona, Spain María Martínez-Adrián Department of English and German Philology and Translation and Interpreting, Faculty of Arts, University of the Basque Country UPV/EHU, Vitoria-Gasteiz, Spain Iris Milán-Maillo Department de Tecnologies de La Informació I Les Comunicacions, Universitat Pompeu Fabra, Barcelona, Spain Laura Panelli Council on International Educational Exchange in Barcelona (CIEE), Barcelona, Spain Elisabet Pladevall-Ballester Departament de Filologia Anglesa Germanística, Universitat Autònoma de Barcelona, Bellaterra, Spain

I

de

Abbreviations

CAH CCLI CEFR CEM CLI CT D/P DCT EFL FFFH FL FTFA GJT GLA HVPT ICLE IFID ILP L1 L2 L2LP L2SFH L3 Ln LPM LT NLI NNS NP O

Contrastive analysis hypothesis Competence CLI Common European Framework of Reference for Languages Cumulative Enhancement Model Cross-linguistic influence Content teacher Declarative/procedural (model) Discourse Completion Test English as a foreign language Failed Functional Feature Hypothesis Foreign language Full Transfer Full Access Hypothesis Grammaticality judgment task Gradual Learning Algorithm High variability phonetic training International Corpus of Learner English Illocutionary force indicating device Interlanguage Pragmatics First language, mother tongue Second language Second Language Perception Model L2 status factor hypothesis Third language Additional language Linguistic Proximity Model Language teacher Native Language Identification Non-native speaker Noun phrase Object

xxiii

xxiv

OT PAM PCLI POS PP QPT S SLA SLM T TLL TOT TPM TT V V2 VLT WC WDCT YLs ZPD

Abbreviations

Optimality Theory Perceptual Assimilation Model Performance CLI Parts of speech Prepositional phrase Quick Placement Test Student Second-language acquisition Speech Learning Model Teacher Third language learning Tip of the tongue Typological Primacy Model Translation task Verb Verb second Vocabulary Level Test Whole class Written discourse completion task Young learners Zone of proximal development

Cross-Linguistic Influence in Second Language Speech: Implications for Learning and Teaching Jaydene Elvin and Paola Escudero

Abstract Cross-linguistic influence (CLI) is a commonly observed phenomenon that influences an individual’s ability to perceive, comprehend and speak in a second language (L2). As phonemic inventories differ across languages and dialects, crosslinguistic difficulty is not uniform. According to the Second Language Linguistic Perception model (L2LP), learners will initially perceive and produce the sounds of the L2 and recognise L2 words in the same way they do so in their own native language. As a result, predictions of L2 difficulty will be learner-specific and dependent on the specific phonetic properties of the native and target language dialects. This chapter presents an up to date and comprehensive description of the L2LP model and the studies that support it. We explain how the model accounts for CLI in L2 speech development, with particular attention to how the model accounts for the effect of L1-based CLI on L2 speech and the differences in learner performance due to dialectal variation. We will conclude by discussing the implications that this model has on language learning and teaching. Specifically, we will highlight the need for L2 teaching to be geared toward a detailed profiling of learners’ native language backgrounds and their individual learning needs and strategies. Keywords Cross-linguistic influence · L2LP model · Individual variation · Dialectal differences · L2 speech perception · L2 spoken word recognition · L2 production

J. Elvin (B) Department of Linguistics, California State University, Fresno, 5245 North Backer Avenue M/S PB92, Fresno, CA 93740, USA e-mail: [email protected] P. Escudero The MARCS Institute for Brain, Behaviour and Development and the Australian Research Council Centre of Excellence for Dynamics of Language, Western Sydney University, Locked Bag 1797, Penrith, NSW 2751, Australia e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_1

1

2

J. Elvin and P. Escudero

1 Introduction Proficiency in second language (L2) speech requires learners to be able to listen, comprehend and speak effectively in the L2. When an L2 is acquired in adulthood, the L1 is already in place and as a result, whether consciously or subconsciously, learners will rely on their knowledge of their L1 to acquire the new language. Given that the L1 and L2 can operate at the same time, it is not unusual to find that the two languages interact with one another. This phenomenon is known as cross-linguistic influence (CLI) and its effects have been observed in many studies of second language acquisition, including L2 speech research. In particular, studies have shown that the differences in the type and number of phonemes in the L1 and L2 may affect a learner’s ability to acquire the sounds of the L2 (e.g., Iverson & Evans, 2007, 2009). Cross-linguistic difficulty is not uniform among speakers of different languages and dialects. This is due to the fact that the similarities and differences between the native and target phonemic inventories influence development in L2 speech acquisition, and consequently, some individuals may find some aspects of the L2 more difficult to acquire than individuals from different L1 backgrounds. The influence of the native language on L2 performance is a core feature of many theories and studies that investigate a learner’s ability to perceive L2 speech sounds. For example, the Speech Learning Model (SLM, Flege, 1995) and the extension of the Perceptual Assimilation Model to PAM-L2 (Best & Tyler, 2007) are two theoretical models of speech perception that claim that L2 sounds are filtered and categorised according to the learner’s native language. In particular, SLM posits that L1 and L2 sounds occur in the same phonological space and sounds which are similar in the L2 to an existing L1 category should be more difficult to acquire than those which do not closely resemble any existing L1 category as the learner must form new categories for those L2 sounds (Flege, 1995). However, PAM-L2 deals with CLI effects by predicting that L2-perceptual difficulties can be explained by articulatory-phonetic similarities between the L1 and L2. The authors claim that we attune to the gesture combinations of our L1 and when the target L2 does not use the same gestures in the combination, learners may have difficulties perceiving those speech sounds. CLI has also been shown to affect learners’ ability to recognise words in speech. The spoken word recognition process can be described as follows: a learner will hear the incoming acoustic signal and a number of possible word forms are activated. The listener must then select one of the activated words as the lexical representation of the spoken word, based on the likelihood of it being correct (Auer, 2009; McQueen, 2007). CLI may affect the number of activated candidates, which results in more competition between words, thus leading to more difficulties and slower spoken word recognition (Vroomen & de Gelder, 1995; Weber & Cutler, 2004). The increased activation and competition in spoken word recognition can be caused by a number of factors resulting from CLI, including (but not limited to) lexical stress (Cutler & Pasveer, 2006; Soto-Faraco, Sebastián, & Cutler, 2001), language experience (Escudero, Broersma, & Simon, 2013), orthographic influences (Hayes-Harb, Nicol, & Barker, 2010; Taft, Castles, Davis, Lazendic, & Nguyen-Hoan, 2008), neigh-

Cross-Linguistic Influence in Second Language Speech …

3

bourhood density (Luce & Pisoni, 1998) and dual activation or similarity between L1 and L2 vocabularies (Dijkstra, Timmermans, & Schriefers, 2000; Dijkstra, Van Jaarsveld, & Brinke, 1998; Weber & Cutler, 2004). Our ability to perceive and recognise speech also has an effect on how we produce sounds in an L2. For example, the Articulatory Settings Theory (Honikman, 1964) investigates cross-linguistic influences and the effect of previously acquired languages on L2 speech production. Honikman (1964) proposes that all languages are characterised by a series of general articulatory positions that involve the lips, tongue, cheeks, jaw and pharynx. Speaking in the L2 may be complicated for learners when the native and target language do not share the same articulatory setting, thus requiring the learner to acquire a new articulatory setting. Furthermore, the Markedness Differential Hypothesis (Eckman, 1977) proposes that target language structures which differ from the L1 and are more implicationally marked will be difficult to acquire in comparison to those which are less marked or considered a universal, namely those structures widely used cross-linguistically (Colantoni, Steele, & Escudero 2015, p. 64; Eckman, 1977). A target language structure that is considered “unmarked” is one that is typical or commonly occurs across a number of different languages, whereas a “marked” language structure is one that is less typical or occurs less among languages. For example, Eckman (2008, p. 96) describes voiced obstruents, oral vowels and open syllables as relatively unmarked in comparison to their marked counterparts—voiceless obstruents, nasalized vowels and closed syllables because the latter language structures occur less frequently across different languages. The aforementioned theories and studies show that CLI indeed affects an individual’s ability to perceive, recognise and produce speech. In this chapter, we will present a review of a theoretical model that accounts for CLI in L2 speech development, namely the Second Language Linguistic Perception model (henceforth L2LP, Escudero, 2005, 2009; Van Leussen, & Escudero, 2015). The main difference between L2LP and the previously discussed models is that L2LP’s foundation is on phonological theory (using Stochastic Optimality Theory: Boersma, 1998) and it accounts for learners at varying developmental stages (beginner, intermediate and advanced competence) and across three speech abilities (perception, spoken word recognition and production). Importantly, the L2LP model focuses on individuals and its descriptions, predictions and explanations are tailored to each learner, which allows for learner profiling at the lab and classroom. We will first provide a detailed explanation of the model and its predictions. We also present a review of the studies that support the L2LP framework and also show the effects of CLI on non-native and L2 speech perception, spoken word recognition and production. We will conclude this chapter by discussing the implications that both the model and the reviewed studies have on language learning and teaching.

4

J. Elvin and P. Escudero

2 The Second Language Linguistic Perception Model—a Framework for L2 Learning and Teaching The Second Language Linguistic Perception model (L2LP; Escudero, 2005, 2009; van Leussen & Escudero, 2015) aims to predict, explain and describe L2 performance in three logical states: the initial state (onset of learning), the developmental state and the end state (ultimate attainment). L2LP is a theoretical and computational model that grew from Boersma’s (1998) Functional Phonology model. The computational implementation of the model was originally conducted using Stochastic Optimality Theory (OT) and the Gradual Learning Algorithm (GLA)1 (Boersma & Escudero, 2008; Escudero, 2005, 2009; Escudero & Boersma, 2004), but has also been implemented using neural-networks (Van Lesussen & Escudero, 2015). Stochastic OT combined with GLA can be used to make very specific predictions regarding individual learning trajectories (Escudero, 2009; Escudero & Boersma, 2004; Yazawa, Whan, Kondo, & Escudero, 2019). The L2LP model contains five key theoretical ingredients for successful L2 speech development (shown in Fig. 1.1) which the author explains are also methodological phases that “allow for a thorough handling of L2 sound perception” (Escudero, 2005, p. 5). The L2LP model thus accounts for individuals from the time they are not yet learners, to the time they start learning the L2 and all the way through to their final stage of L2 learning.

Fig. 1.1 The five ingredients of L2 speech development in the L2LP model

1 In the following sections we provide a brief explanation of the incorporation of Stochastic Optimal-

ity Theory and the Gradual Learning Algorithm in the L2LP. However, for an in-depth discussion of these two concepts in relation to the L2LP model, we recommend that readers refer to Escudero (2009) and Yazawa et al. (2019).

Cross-Linguistic Influence in Second Language Speech …

5

2.1 Ingredients 1 and 2: The Initial L2 State Within the L2LP model, the initial state, also known as the onset of learning, refers to the time at which a learner decides to learn a new language, but has not yet had any training in the target language. Individuals that are in the initial state of learning are often called “naïve” learners of the target L2. To establish the initial state, ingredient 1 of the L2LP model advises to thoroughly describe the optimal perception of each of the languages involved (Escudero, 2009). According to the L2LP model and its optimal perception hypothesis, an ‘optimal listener’ (or native speaker) has a perception grammar which is shaped by the acoustic properties of their production of native sound categories (Escudero, 2009). Therefore, in order to establish the initial state, the acoustic properties of the native and target language need to be described. The optimal perception hypothesis states that the perception of a given speech category is determined by the stochastic ranking of cue constraints that refer to the relevant auditory information. Escudero (2009) explains that optimal L1 perceivers will prefer certain auditory cues that will allow them to differentiate speech sounds (e.g., F1, F2, F3, and duration). And this preference for certain auditory cues (known as cue-weighting) will differ across dialects and languages. Escudero (2009, p. 156) goes on to explain that if two languages differ in the way cues are used to produce L1 speech categories, then “optimal listeners of these languages will have different ways of perceiving these languages”. For example, L1 Japanese speakers show a preference for duration as a cue for differentiating their /i/-/ii/ contrast in perception and production (Yazawa et al., 2019). Likewise, Escudero (2001) found that speakers of two different English dialects used different cues to distinguish the English /i/-/I/ contrasts. In particular, Scottish English listeners relied on F1, whereas Southern British English listeners made use of both F1 and duration to distinguish the vowels. Thus, we see that listeners will differ in their weighting of certain auditory cues to distinguish speech sounds. Therefore, in order to establish the initial state of learning, we must first establish the optimal perception in both the L1 and target language and this can be done by presenting a comprehensive description of the acoustic properties of each language. The second ingredient in the L2LP model proposes that the initial state is a copy of the properties of the native language, known as the full copying hypothesis. In other words, the first thing that a learner will do to begin their process of acquiring the speech sounds of the L2 is to create a duplicate of their L1 perception grammar and assign that to their L2 grammar (van Leussen & Escudero, 2015).2 That is, a learner’s 2 Here

we explain that the initial L2 perceptual grammar is a duplicate copy of the learner’s L1. However, we do acknowledge that there are many students who are bilingual and begin to learn a third language. This may raise the question of which language gets copied? The L1 or L2? Studies in L3 speech perception have found that listeners can selectively choose the vowels of their native (or other) languages that are most relevant to their L3 task. For example, Escudero et al. (2013) suggested that L2 proficiency might influence L3 word recognition if the L2 and the L3 have similar sound inventories. They explain that if learners have learned to use additional cues in the L2 that allow them to distinguish L2 contrasts (e.g., duration and specific formant values), they may be able to use these cues when learning an L3 with a similar sound inventory. Thus it seems that learners

6

J. Elvin and P. Escudero

perception and production of the L2 sounds will look exactly the same as their perception and production of their L1 sounds (see ingredients 1 and 2 in Fig. 1.1). This duplicate thus becomes the L2 perception grammar that gradually modifies itself as it receives perceptual input in the L2, leaving the L1 perceptual grammar intact (Yazawa et al., 2019). A number of studies (e.g., Escudero & Boersma, 2004; Escudero & Chládková, 2010; Escudero, Simon, & Mitterer, 2012; Escudero & Vasiliev, 2011; Escudero & Williams, 2011, 2012; Gilichinskaya & Strange, 2010) have shown through detailed comparisons of acoustic similarity that listeners initially perceive the L2 according to their L1 sounds. For example, Escudero and Vasiliev (2011) tested monolingual Peruvian Spanish listeners’ perception of Canadian French and Canadian English /2/ and /æ/. The authors found that the acoustic similarity between native and target vowels was a very good predictor of context-specific perceptual mapping. Studies such as Escudero, Sisinni and Grimaldi (2014) also show that learners whose languages share similar vowel inventories (such as Salento Italian and Peruvian Spanish) perceive the target L2 differently. This is because the acoustic realisations of vowels will differ in each language, despite the shared vowel inventory. Thus, the copy of their L1 perception grammar that learners make to perceive the L2 will differ, resulting in differences in non-native vowel perception between the two groups. Most previous studies focused on L2 perception but more recent evidence further shows that individuals will initially produce L2 sounds in the same way as they produce acoustically close L1 sounds. In Elvin, Williams and Escudero (2016), eight European Spanish monolinguals were asked to perceive and produce six Brazilian Portuguese (BP) contrasts (/a/-/2/, /a/-/O/, i/-/e/, /e/-/2/, /o/-/O/, /o/-/u/). The authors successfully predicted that the participants would have difficulties producing stable categories for the non-native vowel contrasts that were perceptually difficult to perceive (e.g., /i/-/e/ and /o/-/u). The vowel contrasts that were perceptually easy to discriminate (e.g., /a/-/2/) were produced as separate vowel categories in novel BP words, however these separate categories were still heavily influenced by their L1. A similar study was conducted in Elvin, Escudero, Williams and Best (2016), but with monolingual Australian English speakers. The authors had the speakers discriminate and categorise BP vowels and used those results to predict their non-native production. As with the aforementioned study on Spanish speakers, the Australian English participants also had difficulties producing separate vowel categories for those BP contrasts that were difficult to discriminate. Likewise, their production of BP closely resembled the acoustic properties of Australian English. An example of one male Australian speaker’s non-native BP vowel production compared against the target BP vowels and his own native vowel production is shown in Fig. 1.2. In this figure we can see that the speaker’s F1 and F2 values of their non-native BP vowels (in blue) fall in the area of his native Australian vowel productions (in black), and are quite different from the target BP vowels (green, with circles). In will rely on the language that has similar acoustic properties to the target language when learning L3 speech sounds. However, we do acknowledge that the L2LP implementation of the selective use of sound inventories with the respective perception grammar is yet to be investigated.

Cross-Linguistic Influence in Second Language Speech …

7

Fig. 1.2 A sample of one male participant’s F1 and F2 values for all their non-native Brazilian Portuguese vowel tokens (blue), displayed with the mean values of their native Australian English vowel productions (black) and the average values for the target BP vowels (black, with circles)—this figure is based on results reported in Elvin, Williams, et al. (2016)

particular, the BP high and mid vowel contrasts are merged and the low vowel has higher F1 values than those of the target vowel. Based on these results, we can see that learners indeed use a copy of their own native vowel production in their initial L2 production. The L2LP’s full copying hypothesis has important implications for the learner’s future L2 speech development as it determines the individual’s learning task (ingredient 3) within their L2 development (ingredient 4) as will be discussed below. In the L2LP model, the best method for establishing the initial state (i.e., the full copy of the learner’s L1) and predicting their learning task is to use Stochastic OT as well as comprehensive acoustic analyses of the native and target languages and a comparison between the two. The use of Stochastic OT with comprehensive acoustic analyses of both the L1 and L2 have been very successful in predicting L2 difficulty, more so than machine learning algorithms such as Naïve Bayes and Nearest Neighbour (see Escudero and Bion (2007) and Yazawa et al. (2019) for a discussion). Some studies (e.g., Escudero, 2009; Escudero & Boersma, 2004; Van Leussen & Escudero, 2015) have also been successful in using the acoustic data from the L1 and L2 to predict L2 development computationally. Other studies using the L2LP framework (e.g., Elvin, Williams, & Escudero, 2016; Escudero & Vasiliev, 2011) have used other statistical analyses such as discriminant analyses to predict cross-linguistic acoustic similarity and differences between the native and target language. In this method, a statistical model is trained on the native vowel production data with input parameters (e.g., F1, F2, F3 and duration) that are used to identify vowels in the L1. To determine crosslinguistic acoustic similarity, the model then tests the L2 vowel production data and classifies the L2 vowel tokens as L1 categories based on those input parameters. In other words, the model will examine the acoustic properties (formant and duration values) of the L2 token and match it with the L1 category that is the most similar. Thus, comprehensive acoustic analyses are essential for making predictions regard-

8

J. Elvin and P. Escudero

ing L2 development when using methods such as Stochastic OT, statistical analyses and computational modelling.

2.2 Ingredient 3: The Learning Task The learning task in the L2LP model is directly connected with the initial state of leaning as the cross-linguistic acoustic comparisons of the L1 and L2 can be used to predict the individual’s learning task. Recall that in the initial state a learner will make a duplicate copy of their L1 and assigns that copy to their L2. The learning task therefore relates to how much this L2 duplicate copy needs to develop and change in order to match the target L2 perception and production. The L2LP model therefore puts forward three possible “learning scenarios” that may emerge at this stage, namely the NEW, SIMILAR and SUBSET scenarios. The NEW scenario occurs when two non-native or L2 sounds are acoustically similar to, and subsequently perceived and categorised as one single native sound. This scenario creates a difficult learning task for the learner as they will need to either create a new sound category or split their existing L1 category. The SIMILAR scenario occurs when two non-native sounds are acoustically similar to, and perceived and categorised as two separate native categories. The learning task in this scenario is predicted to be much easier as learners will be able to replicate their existing L1 categories and adjust their perceptual boundaries as needed. The third scenario occurs when two non-native or L2 sounds are acoustically similar to and/or perceived and categorised as two or more native L1 categories, known as the SUBSET scenario. The degree of perceptual difficulty in this scenario depends on whether or not an acoustic and/or perceptual overlap occurs. If the two non-native or L2 sounds are acoustically similar or perceived as the same native categories, then an acoustic and/or perceptual overlap is said to have occurred, resulting in a neutralisation of the L2 contrast (and is therefore comparable to the NEW scenario). But if the two sounds are perceived as separate multiple categories then no acoustic and/or perceptual overlap has occurred and no difficulties are expected. Learners may face a particular challenge in this scenario, known as the SUBSET problem. This problem occurs when a certain contrast does not exist in the target language and the learner must stop perceiving differences that are not meaningful in the target language, a task that may pose considerable difficulty, especially when a perceptual overlap has occurred. A visual representation of these learning scenarios is provided in Fig. 1.3. It is important to note here that these L2LP learning scenarios should not be confused with the description of the perceptual assimilation patterns in PAM-L2 (Best & Tyler, 2007) or equivalence classification as described in SLM (Flege, 1995). Firstly, SLM is primarily concerned with the similarity and dissimilarity between target L2 sounds and individual sound categories in the native language, whereas PAM-L2 and the L2LP model are concerned with the mapping of the two target speech sounds in an L2 contrast to the learner’s own L1 categories. Secondly, although L2LP’s NEW and SIMILAR scenarios are indeed comparable to PAM-L2’s single and two-category

Cross-Linguistic Influence in Second Language Speech …

9

Fig. 1.3 A visual representation of the L2LP learning scenarios

assimilation patterns, the two models differ in their explanation and use. Specifically, in PAM-L2 the patterns of assimilation are only used to predict difficulties in nonnative and L2 discrimination (and in some more recent studies, word recognition). The L2LP model differs from both SLM and PAM-L2 in that it “proposes precise learning tasks and developmental trajectories for learners, depending on the learning scenario with which they are confronted” (Van Leussen & Escudero, 2015, p. 2). For example, as described above, in L2LP’s NEW scenario where two L2 sounds are perceived as a single L1 category, the prediction is that the contrast will be difficult to perceive, recognise in a word and produce and the resulting learning task is a difficult one where the learner must either split their L1 category or create a new category entirely in order to acquire that sound contrast. Finally, the L2LP model’s use of the terms “NEW” and “SIMILAR” differ to their use in SLM. For example, in the SLM, a “new” category is formed when the L2 phoneme does not closely resemble any phonetic/phonemic category in the L1 and the SLM predicts that learners should find this “new” category much easier to acquire than an L2 phoneme that is “similar” to an existing L1 category which despite phonetic differences should be more difficult to acquire. Conversely, in L2LP, the learning task resulting from the SIMILAR scenario is considered easier than those in the NEW scenario because in the former the learner does not need to form any new category but simply adjust their perceptual boundary between categories. There are a number of studies that have tested L2LP’s predictions for these learning scenarios. For example, Escudero and Williams (2011) investigated Peruvian Spanish listeners’ categorisation of Dutch vowels in order to predict their L2 development. The authors found that their cross-linguistic acoustic comparisons of the native and target language generally predicted their listeners’ non-native categorisation patterns. Given that the Dutch vowel inventory is larger than the Spanish vowel inventory, they successfully predicted many cases of L2LP’s NEW scenario (where two nonnative vowels are perceived as a single native category). Based on these results, the authors suggest that in order to accurately perceive Dutch vowels, Spanish learners will likely be faced with the learning task of splitting their existing L1 categories.

10

J. Elvin and P. Escudero

L2LP’s learning scenarios were also investigated in Elvin, Escudero and Vasiliev (2014) who tested Australian English and Iberian Spanish listeners’ perception of Brazilian Portuguese vowels. The authors found that although the two languages differ in the size of their vowel inventory, acoustically the vowels fall in and around the same locations in the acoustic space when compared to Brazilian Portuguese. The authors therefore predicted and indeed found that both groups of listeners experienced the same contrasts easy and or difficult to perceive. However, the authors did find that overall the Spanish group performed better on the discrimination task and the authors hypothesised that the Australian participants may have had more difficulties because they have a larger vowel inventory which may have caused them to activate more competing vowel categories than the Spanish participants who have fewer vowels in their inventory (and as a result, fewer competing categories). The authors further suggest that vowels which are acoustically closer to the target vowels with no activation of competing vowel categories are easier to discriminate.

2.3 Ingredient 4: L2 Development In the L2LP framework there are two stages of L2 development: (1) perceptual learning (using the Gradual Learning Algorithm) and (2) recognition learning. Recall that at the initial state of learning, a learner will create a duplicate copy of their L1 perceptual grammar for their L2 perceptual grammar. The L2LP model proposes that a learner’s Gradual Learning Algorithm (GLA3 ; Boersma, Escudero, & Hayes, 2003) is responsible for changes in their L2 perceptual grammar to closely resemble that of a native speaker of the target language. According to Escudero (2009), the GLA learns to classify speech sounds via distributional learning, a human learning mechanism whereby perception is affected by the phonetic distribution of speech sounds along an acoustic continuum that encompass the two sound categories (Escudero, Benders, & Wanrooij, 2011; Maye, Werker, & Gerken, 2002). In fact, distributional learning is considered to be the learning mechanism that underlies the acquisition of sound categories in the first year of life (Wanrooij, Escudero, & Raijmakers, 2013). In the L2LP framework, distributional learning is the learning mechanism that leads the creation of new sound categories in the L2 and has been shown to have immediate and long term effects in L2 perception (see Escudero & Williams, 2014). Perceptual learning using the GLA is also error-driven in that learners will attempt to improve their perception or adjust their use of auditory cues by updating their perception grammar when they mishear a sound. Recall that at the initial state of learning, L2 learners have already established their optimal perception in the L1 using Stochastic OT (ranking constraints or cue-weighting). When listening to incoming speech sounds, learners will categorise the L2 sounds according to the distribution of L1 categories based on those ranked auditory cues (e.g., spectral cues 3 See

Escudero (2009), Escudero and Boersma (2004) and Yazawa et al. (2019) for full details on how the L2LP in combination with the GLA explain L2 perceptual learning.

Cross-Linguistic Influence in Second Language Speech …

11

and duration). However, when there is a mismatch between the word the speaker intended and the word they heard, the learner notices this error and gradually shifts or adjusts their native cue-weightings to match those of the target language in order to lower the probability that the same error will occur again. Through the GLA two perceptual learning mechanisms, a learner can develop new L2 sound categories (distributional learning, e.g., duration for Spanish learners of English) or decrease their use of a particular cue when learning a contrast with similar sounds in the L1 and L2 (error-driven learning, e.g., English and Japanese listeners need to adjust their cue-weighting of the durational acoustic cue in order to perceive the French /ae/-/2/ contrast (English learners of French, Escudero, 2009), and the English /i/-/I/ contrast (Japanese learners of English, Yazawa et al., 2019). The L2LP model also proposes a direct link between speech perception and spoken word learning and recognition, in that phonetic contrasts which are confused in speech perception are often also confused in spoken word recognition. Specifically, learning to recognise words in a new language is message-driven (or meaning-driven in the revised L2LP; Van Leussen & Escudero, 2015) because learners have to infer words based on how well they are able to understand the meaning intended by the speaker. The L2LP model’s hypotheses for message-driven learning has been tested computationally in Van Leussen and Escudero (2015) where simulated learners showed progress toward native-like perception and recognition of Dutch front vowels and adapted their L2 in a similar manner to real-life L2 learners. Thus, according to the L2LP model a learner improves their L2 perception by gradually shifting their cue-weighting so that it resembles cue-weighting in the L2. This gradual shift is achieved through error-driven learning (whenever a misunderstanding has occurred) and distributional learning. On the other hand, L2 speech recognition is developed through message-driven (or meaning-driven) learning because the message (or semantic context) triggers learning if there is an error or misunderstanding in the interpretation, while maintaining their L2 perception grammar. The L2LP model also accounts for two processing strategies in this stage of L2 development. In particular, lexical processing can either be strictly sequential (i.e., bottom-up, as previously described in the original version of L2LP) or interactive where lexical feedback to perception is allowed. Van Leussen and Escudero (2015) explain that sequential processing involves a two-step process: The sequential learner first processes the acoustic to phonetic connections or mappings of perception. After that connection has been processed, the sequential learner then processes the phonetic to phonemic as well as the phonemic to lexical connections that lead to subsequent recognition of the spoken word. Interactive processing differs from sequential processing in the fact that all of these connections of recognition (acoustic—phonetic; phonetic—phonemic; phonemic—lexical) are processed and evaluated together (i.e., at the same time). These two processing strategies were tested computationally in Van Leussen and Escudero (2015) where they found that they ultimately led to correct L2 recognition, but the strategies differed in regards to the phonetic representations and amount of exposure required to acquire the L2 contrast. Specifically, the simulated learners using a sequential processing strategy needed a larger amount of input data (exposure to L2 words) than learners using an interactive processing strategy

12

J. Elvin and P. Escudero

to achieve the same level of success with the L2 contrast. To date, these two processing strategies have only been tested through computational modelling, however, some evidence for these different processing strategies may be found in Elvin (2016). Specifically, in a word learning and word recognition study, Elvin (2016) found that eight listeners (5 Australian, 3 Spanish) had lower accuracy scores in their recognition of Brazilian Portuguese minimal pairs that contained a perceptually difficult vowel contrast, yet had comparable scores for perceptually easy contrasts. The author hypothesised that these eight learners might be sequential learners who require additional exposure in order to be able to recognise word-pairings containing difficult vowel contrasts. However, further investigation is required to be able to successfully identify sequential learners and to determine whether additional exposure to the L2 does indeed lead to improved performance for these individuals.

2.4 Ingredient 5: The End State The final ingredient in the L2LP model is known as the end state or ultimate attainment. Escudero (2005) uses the term “end state” to describe the varying L2 proficiency levels that learners reach that may or may not match the target. According to the L2LP model, both the speed at which one reaches this final state, as well as the specific developmental path, will differ across individuals depending on the specific L2 learning task that the leaner is required to undertake (i.e., NEW, SIMILAR and SUBSET). Furthermore, ultimate attainment in the L2 will depend on whether or not the learner receives rich L2 input over the course of L2 development. Specifically, the more L2 input the learner receives during the developmental stage, the more likely they will achieve optimal perception (and production) in the L2. Importantly, the model hypothesises that by the time the learner reaches the end state they will have separate grammars and language activation modes that will allow them to achieve optimal perception of the L2 and at the same time preserve their native L1 perception (Escudero, 2005, p. 327). An individual’s L1 perception will only be affected by the L2 if they are not exposed to sufficient L1 input. Thus in order to attain optimal L2 perception (and production) and maintain optimal L1 perception and production, the learner must be exposed to rich production input in both languages. Again, the end state in the L2LP model has only been tested in computational modelling. For example, Boersma and Escudero (2008) found that after 18 years of “simulated” input, a learner was able to achieve an end state of 75% recognition accuracy. In Van Leussen and Escudero (2015), simulated sequential and interactive learners reached a stable recognition of around 85% accuracy, however, the simulated sequential learners required more input in order to achieve the same end state as the interactive learners. Finally, Escudero (2009) and Yazawa et al. (2019) showed that their simulated learners gradually shift their cue-weighting, with the final stage of learning being very similar to a simulated native listener, while keeping their L1 perception grammar intact. As for evidence supporting the L2LP perception modes hypothesis, Escudero (2009) shows that the same French tokens were classified dif-

Cross-Linguistic Influence in Second Language Speech …

13

ferently depending on whether the learners were in a French or English session and that French proficiency predicted accurate classification. In Yazawa et al., the L2LP’s language mode hypothesis was further tested on real Japanese learners of English and showed that the learners relied on different cues (duration or spectral) depending on the language session they were in (English or Japanese), despite the stimuli remaining the same. To summarise the proposal, the L2LP model assumes that all learners will go through three stages in their L2 speech acquisition, namely initial, developmental and end state. According to the L2LP framework, in order to predict L2 development, we must establish the optimal perception of each language using the best possible framework to account for a fine-detailed analysis of how auditory cues are weighed (e.g., Stochastic OT seems quite successful but neuro-networks work better for some cases) as it allows us to have a precise understanding of the perception grammars of the native and target languages. Then, learners will start their development by creating a duplicate of their L1 perception grammar for the L2. A detailed acoustic comparison of the similarities and differences between the L1 and L2 can provide information regarding the specific learning task the individual must overcome during the developmental stage in order to attain native like perception and production of the L2. In the developmental stage, the GLA, which implements distributional learning and error-driven learning, is used to explain how perceptual learning occurs as learners gradually change their native cue-weightings to match those of the target language. Learners also develop their spoken word recognition through meaning-driven (or message-driven) learning as they attempt to update their lexical representations based on the semantic context. At the final state of learning, individuals will have developed a separate L2 grammar, while their L1 grammar remains intact, thus allowing them to switch between two different language modes that are optimal for each of their languages. It is important to note that according to the L2LP framework, the end state will differ depending on each individual, according to the amount of L2 input they receive. If learners receive rich L2 input, they should be able to attain optimal L2 perception and production.

3 L2LP: Implications for Teaching and Learning In the previous section we have provided a detailed description of the L2LP framework and have shown how it can predict, describe and explain the developmental path of individual learners from the initial until the end state of L2 learning. Importantly, the L2LP framework can be used as a tool in L2 teaching (specifically in the development of language software or language learning applications) to profile individual learners and predict their development, as will be discussed in Sect. 3.3. Furthermore, the model accounts for the cross-linguistic influences that may influence a learner’s developmental patterns. For example, in our detailed account of the L2LP model, we have shown the model’s take on CLI through explaining that an individual’s L1 will shape L2 acquisition from the onset. Specifically, at the initial

14

J. Elvin and P. Escudero

state a learner’s L2 perception and production will match perception and production in the L1. However, the model also accounts for CLI influences stemming from a specific target L2 dialect, which we will discuss below together with the role of individual variation within the model. We will conclude this chapter by describing how the L2LP model may be further applied to actual language learning settings.

3.1 L2LP and the Effect of Dialect Recall that the L2LP model states that in the initial state of learning, an individual will make a duplicate copy of their L1 perception and production for their L2 perception and production. The assumption is therefore that speakers of the same native language, but with different dialects, will differ in their L2 developmental paths. According to the L2LP model, the acoustic differences in L1 vowel production in each dialect will result in differences in perception, spoken word recognition and production. For example, Escudero and Williams (2012) tested Peruvian and Iberian Spanish learners of Dutch and found that despite the fact that both dialects share the same native vowel inventory their acoustic realisations differed and as a result the acoustic similarity between each dialect and Dutch differed. In particular, the authors found that the vowels in the Dutch /a/-/A/ were acoustically similar to the same native category in Peruvian Spanish, but separate native categories in Iberian Spanish. The authors further found that the Iberian Spanish listeners’ performed better in their discrimination of the Dutch /a/-/A/ contrast than the Peruvian Spanish listeners. Chládková and Podlipský (2011) also found that Bohemian and Moravian Czech listeners differed in their categorisation of Dutch, suggesting that even when the two dialects differ only very slightly, learners may follow different developmental paths in their acquisition of the L2. Studies have also shown that learners differ in their perception of two different dialects in the target language. For example, Escudero and Boersma (2004) found that native Spanish learners of English behave differently when acquiring the English sheep-ship contrast depending on whether their target dialect was Scottish English or British English. In particular, those whose target dialect was Scottish behaved in a similar manner as native Scottish speakers, however, those whose target dialect was Southern British behaved in a manner that was distinct from both native Scottish and native Spanish patterns. Escudero and Chládková (2010) tested Peruvian Spanish listeners’ categorisation of American English and Standard Southern British English and showed different categorisation patterns by the same learners for each dialect. The fact that learners differed in their categorisation of American and British English vowels leads to the prediction that not only their initial state of learning will be different, but their L2 perceptual development and specific learning task will also differ. These findings relating to dialectal effects on L2 speech perception have important implications for language teaching and learning. Language teachers may benefit from the knowledge of the relationship between their learners’ native dialect and the target

Cross-Linguistic Influence in Second Language Speech …

15

language. For example, at the beginning of their course, teachers might choose to investigate the language backgrounds of their students and the relationship between native L1 sound system and the target L2 system that is being taught. Teachers should also be careful not to assume that because students may share the same native language, they will have the same problems when speaking in the L2. The aforementioned studies have shown differences in L2 perceptual patterns for learners who share the same native language, but differ in their dialect and it is important that teachers cater to these dialectal differences. Likewise, we have discussed that the same learners will differ in their perception depending on which L2 dialect they are exposed to. Therefore, learning and teaching might be more effective if it is dialect specific or if students are exposed to a range of different dialects and are trained to perceive those dialectal differences.

3.2 Individual Variation Within the L2LP Model It is commonly accepted that learners will vary in terms of how successful they are in learning a new language. Learners differ in relation to the strength of their foreign accent, the speed of their L2 acquisition, and their ultimate level of achievement (the end state) also varies across individuals (Ellis, 2004). The growing body of literature has found age of acquisition, language exposure, language aptitude, motivation, cognitive abilities and differing learner strategies and styles (e.g., Aliaga-Garcia, Mora, & Cerviño-Povedano, 2010; Dörnyei & Shekhan, 2003; Ellis, 2004; Perrachione, Lee, Ha, & Wong, 2011) to be key factors that explain individual variation in L2 learning. For example, Perrachione et al. (2011) found that the ability to learn a phonological contrast for pitch depended on the interaction between individual differences in perceptual abilities and the design of the training program. In other words, they found that learning achievement was dependent on their pre-training aptitudes. The authors found that individuals with weaker pitch perception abilities were disproportionately impaired in a high-variability training environment in which perceptually stronger learners excelled. In another study, Aliaga-Garcia et al. (2010) showed that cognitive abilities such as phonological short-term memory plays a role in an individual’s ability to form L2 phonetic categories. In particular, the authors showed that advanced EFL learners classified as having a high capacity for phonological short-term memory had higher accuracy scores (following training) in vowel identification and discrimination than the advanced EFL learners with a lower capacity for phonological short-term memory. As previously mentioned, the L2LP model was designed to account for individual variation. The framework focuses specifically on individual learners in that the model’s descriptions, predictions and explanations are tailor-made to each learner, which allows for learner profiling at the lab and classroom. There are a number of studies that use the L2LP theoretical framework to investigate individual differences (e.g., Escudero & Boersma, 2004; Mayr & Escudero, 2010; Wanrooij, Escudero, & Raijmakers, 2013). For example, Escudero, Benders and Lipski (2009) investigated

16

J. Elvin and P. Escudero

the use of vowel spectrum (vowel height, backness and lip rounding) and duration as cues for the categorisation of Dutch vowels by three groups of listeners: L1 Dutch, L1 Spanish learners of L2 Dutch and L1 German listeners. The results indicated that experience with the language indeed helps categorisation of the L2 vowels as the L1 Dutch and Spanish learners of Dutch had more reliable categorisation of the Dutch /a -A/ vowel contrast than the L1 German listeners with no experience with Dutch. Furthermore, the authors found a large amount of individual variation in cue weighting across the three groups. Specifically, the L1 Dutch and L1 German groups weighted the spectral cues heavier than the duration cues, whereas the Spanish learners of Dutch weighted duration cues more heavily. Interestingly, 14 of the 38 Spanish learners of Dutch weighted spectral cues heavier than duration cues. The authors posit that some non-native speakers rely on spectrum (vowel qualities) rather than duration when categorising non-native vowels and that future investigations with larger participant numbers may highlight different cue weighting strategies among individual learners. Perhaps one of the most comprehensive accounts of individual variation according to the L2LP theoretical framework is that of Mayr and Escudero (2010). The authors investigated whether native English first or fourth year undergraduate learners of German follow different developmental learning paths in their perceptual development. The study involved two experiments, the first one collected perceptual assimilation patterns of L2 vowels which were subsequently used to predict leaner’s’ performance in an L2 identification task. The results indicated that despite some inconsistencies, individual performance in the L2 identification task was largely consistent with the predictions made based on their individual perceptual assimilation patterns. Mayr and Escudero (2010) further found that the variability in the perceptual assimilation patterns was systematically constrained. That is, despite variation in terms of exact classification percentages, native English listeners were consistent in selecting only a subset of the 13 native response options available to them when assimilating the six German vowels to native vowel categories. Additionally, Wanrooij et al. (2013) demonstrated that distributional vowel training (see Sect. 2.3 of this chapter) can help individual learners to improve their classification of difficult non-native contrasts. Interestingly, the findings revealed that the changes in the use of perceptual cues after training were related to an individual’s listening strategy (i.e., a specific way of using acoustic cues) prior to training. Participants were given a pre-training discrimination task and were divided into “high performers”, if they were able to use F1 (vowel height) and F2 (vowel backness) cues to discriminate the target vowels with higher accuracy scores, and “low performers” if they had lower accuracy scores and did not focus on the combination of F1 and F2. After training, those classified as “low performers” started to make use of F1 and/or F2 and those originally classified as “high performers” added a subtle secondary cue, namely F3 which corresponds to lip rounding. The above findings all support the L2LP claim that individuals differ in terms of their predicted L2 developmental sequences (from the onset of their learning) even when common factors that influence individual variation are controlled (e.g., age, dialect, language experience, etc.). Specifically, learners will follow different  

Cross-Linguistic Influence in Second Language Speech …

17

developmental paths depending on their individual categorization patterns and the learning scenarios (NEW, SIMILAR, SUBSET) that are identified. For example, for a given L2 contrast, the cross-linguistic acoustic comparison may identify a NEW scenario for one learner and a SIMILAR scenario for another. These two learners would therefore follow different developmental paths. For the learner facing the NEW scenario, the learning task would be one where they either split a native L1 category or create a new category in order to acquire that sound contrast, whereas the learner who faces a SIMILAR scenario need only replicate their existing L1 categories and adjust their perceptual boundaries in order to develop this contrast. Furthermore, as described above, an individual’s ability to use certain acoustic cues to perceive differences in vowel contrasts will also influence the speed in which they acquire specific L2 contrasts. For example, in the aforementioned study by Wanrooij et al. (2013), the authors found that participants who were able to make use of F1 and F2 to distinguish contrasts performed better after distributional vowel training, as they had acquired and were able to make use of an additional cue, namely F3. We therefore predict that an individual who is able to use a combination of acoustic features (duration, F1, F2 and F3) to distinguish between speech sounds is likely to acquire that L2 contrasts more quickly than an individual who is not able to use all of those cues. To further explore the L2 developmental path, Escudero et al., are currently working on a longitudinal study that investigates the acquisition of contrasts over the course of one and a half years.

3.3 The L2LP Model—A Framework for Education To date, the majority of the studies that investigate L2 perception, spoken word recognition and production have been conducted in a laboratory. While these studies contribute a great deal of knowledge to the field of L2 acquisition, most are published in journals that are not teacher-oriented (Derwing & Munro, 2005). As a result, there seems to be a lack of transfer of knowledge from research into L2 speech acquisition, particularly in the area of pronunciation, and teaching pedagogy (Levis, 2016). The fact that most L2 speech research is targeted toward a specialised audience may be a contributing factor to the lack of uptake in phonetic training (both in perception and production) in L2 classrooms.4 Additionally, the lack of training for teachers in this specialized area has possibly resulted in a lack of confidence and willingness to teach pronunciation (Levis & Wu, 2018). Earlier, we claimed that the L2LP model is a good candidate as a framework for education, however as with most research in L2 speech, most of the journals in which research using the L2LP framework appear is not teacher-oriented. Furthermore, there is a lack of classroom-based research in 4 The

fact that the CATESOL journal (2018) recently published a special issue on Pronunciation, with many of its articles focusing on how pronunciation can be taught in the classroom highlights the growing importance of establishing a connection between L2 speech research and teaching pedagogy.

18

J. Elvin and P. Escudero

support of the L2LP model. For this reason, in this section we discuss how the model can be used as a tool in L2 teaching and highlight a need for the development of both tools and research investigating its applicability in real-world settings. The L2LP framework can be used as a tool in L2 teaching to profile individual learners, identify the specific L2 difficulties that they may have as a result of crosslinguistic influences, and predict their further development. The model is designed to predict, explain and describe L2 speech acquisition (in perception, spoken word recognition and production) from the initial state (or onset of learning) up until ultimate attainment for individual learners. One of the model’s core assumptions is that cross-linguistic influences between the native and target dialect affect the learner’s developmental patterns. According to the L2LP framework, learners will initially perceive and produce L2 sounds in the same way as they perceive and produce their own L1 sounds. That is, they make a duplicate copy of their L1 perception grammar that gets updated as they continue learning the L2. Strong emphasis is placed on acoustically comparing the cross-linguistic similarity between the L1 and target L2 dialect, in addition to in the cross-linguistic acoustic similarities/differences in order to determine the learning scenarios as well as establishing their categorisation of the L2 sounds as native sounds in order to identify the specific learning scenarios (NEW, SIMILAR and SUBSET) that they will need to overcome in order to acquire specific L2 contrasts. Such knowledge would be beneficial to language teachers who are training their students in L2 pronunciation. According to the L2LP model, the first step in predicting L2 speech development is to establish the initial state. Teachers may benefit from a description of the acoustic similarities and differences between their students L1 and the target L2 dialect. They may also consider obtaining L2 categorisation patterns from their students in order to identify the specific L2 contrasts that learners will struggle with and use that information to train them to improve their perception and production of those sounds. This is important because our ability to produce speech sounds depends on our ability to perceive them, yet very few pronunciation (and even listening) classes focus on L2 perception because the focus has commonly been on the training of the appropriate use of the speech articulators to produce the L2. For instance, in Yoshida’s (2018) review of the technological tools that can be implemented in language programs teaching pronunciation, the majority of the tools are designed to either model pronunciation of speech sounds by having learners listen to the sound, see a diagram of how the sound is produced and see a video of a speaker producing that sound, or the tools are targeted toward pronunciation practice through the use of narration using visual aids (see Yoshida, 2018 for a full review). It seems that there are very few resources available that encourage learners to train their perception, despite its influence on pronunciation. For example, a number of studies have shown that high variability phonetic training (HVPT) leads to increased ability to perceive and produce L2 sounds (see Barriuso & Hays-Harb, 2018 for an excellent review). HVPT is a method whereby listeners are trained to listen to speech sounds presented by multiple speakers rather than one speaker. The training usually consists of identification tasks (typically two-alternate forced choice, see Logan, Lively and Pisoni (1991) or Wang and Munro (1999) for examples) with or without corrective feed-

Cross-Linguistic Influence in Second Language Speech …

19

back. Acknowledging the lack of phonetic training in language teaching contexts, Thomson (2012) identified HVPT as a good candidate for computer assisted pronunciation teaching and developed the English Accent Coach website (Thomson, 2012), which was designed to complement L2 pronunciation teaching. Learners using the program are exposed to HVPT and are trained in their perceptual abilities as they play a game that involves the identification of speech sounds or a game in which they need to recall a sequence of sounds that they hear. Thomson and Derwing (2016) empirically tested the effectiveness of the English Accent Coach website and the results indicated that the targeted phonetic training did indeed lead to improvement in L2 pronunciation. Earlier, we explained that according the L2LP model, individuals acquire speech sounds through distributional learning and error-driven learning and mentioned that studies (e.g., Escudero & Williams, 2014) have shown that exposure to frequency distributions can lead to improvement even after 2 min of exposure in perceptual training. Just as the HVPT was turned into a computer assisted pronunciation teaching game, it may be worthwhile to apply a similar procedure that could be applied for distributional learning. That is, distributional learning tasks could be developed into online games for websites and mobile applications that train learners to perceive and ultimately produce differences in L2 speech contrasts. In fact, Escudero et al., are currently developing a language exposure app that trains parents and their infants to improve their speech processing abilities in order to learn foreign language words. We are also currently brainstorming other possibilities for mobile applications or websites using the L2LP framework that can be used to identify the L2LP learning scenarios in individual’s categorization patterns and train those learners in the perception, spoken word recognition and production of those speech sounds. If such educational software could be developed, future research would of course need to investigate the effectiveness of the language tools for individual learners as well as for L2 pronunciation teachers. The L2LP model also makes predictions regarding the two processing strategies that learners use to recognise spoken words and the identification of these processing strategies might be beneficial in a classroom setting. Recall that L2LP explains that learners will either process information sequentially (i.e., bottom up processing) or interactively (allowing lexical feedback to perception). Given that the amount of exposure needed to learn new words differs depending on the learner’s processing strategies, it would be worthwhile to identify each learner’s strategy prior to instruction so as to effectively cater to their individual word learning needs. As previously stated, these processing strategies are yet to be tested on real learners and future studies should consider testing them in a language classroom. For example, it would be useful to devise a tool that can be used by teachers to identify an individual’s processing strategy to help them identify the best approach for teaching their students. If certain learners are identified as sequential learners, then teachers will know how to provide the extra input that these leaner’s need in order to recognise spoken words.

20

J. Elvin and P. Escudero

4 Summary and Conclusion In this chapter we have presented an in-depth account of the L2LP model of speech development which can be used to predict, describe and explain speech acquisition in perception, spoken word recognition and production. We have presented the model’s claims for acquisition at the initial, development and end state of learning and have discussed the studies that provide evidence of these claims. We have also discussed the implications of cross-linguistic influences and in particular this learning model in L2 teaching and learning. In the L2LP model, these cross-linguistic influences manifest themselves in the sense that perception and production of the L2 is filtered through the L1 and that individual learners will show different performance patterns based on the dialect that they speak and the dialect that they are learning. We have also suggested some practical applications of the L2LP model in language teaching and learning. For example, distributional learning may be incorporated into language classrooms or mobile applications for language learning. A tool that can be used to identify individual learning strategies (sequential vs. interactive) may also be useful for teachers. While the application of the L2LP model as a framework for language education and a tool for learner profiling may indeed be very beneficial to language teachers and learners alike, software would first need to be developed and future studies should be conducted that adequately investigate such implementation in realword teaching situations.

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/*. A Preliminary Study on the Nature and Impact of Phonological and Orthographic Transfer in the English Speech of Bilingual Speakers of Spanish and Galician Yolanda Joy Calvo-Benzies Abstract The present paper aims to describe instances of phonological and orthographic interference in the English speech of bilingual speakers of Galician and Spanish. Data were collected from both high school and university level subjects, and three spoken tasks were used: (a) a picture-story; (b) a text reading; and, (c) oral presentations. General British was taken as a model in classifying mistakes. Findings reveal that a high number of the pronunciation mistakes made by bilingual speakers of Spanish and Galician when speaking in English can be attributed to phonological or orthographic interference from their native languages. More particularly, at the phonological level, these subjects had identifiable problems with /N/, word stress, and certain initial and word-final consonant clusters. Furthermore, many mistakes due to orthographic transfer were registered across all groups of participants and tasks, most of which could be categorised into two groups: (a) transfer involving the pronunciation of a Spanish vowel guided by the spelling of a word; and (b) interference in which speakers did not pronounce unstressed syllables with a schwa in spellings like but rather with a full Spanish vowel followed by a /R/ sound. Some pedagogical implications of learner findings will be discussed in the last section. Keywords Pronunciation · Orthography · Transfer · Bilinguals

1 Introduction When learning a new language, it is very likely that a learner will encounter a number of differences between the phonological and phonetic systems of their First Language(s) (L1s) and the Foreign/Second Language (FL/L2) in question. A common process, employed extensively by many learners to overcome these difficulties, is Y. J. Calvo-Benzies (B) Departamento de Filología Española, Moderna y Clásica, University of the Balearic Islands, Carretera de Valledemossa Km 7.5, 07122 Palma de Mallorca, Spain e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_2

21

22

Y. J. Calvo-Benzies

that of “the mechanism of category assimilation when category formation has been blocked”; hence, “the L2 learner may initially produce the L2 speech sound just as if it were the corresponding L1 speech sound, that is, without any modification” (Flege, 2007, p. 353). In broad terms, when one learns a new language, the L1(s) “phonetic systems reorganize in response to sounds encountered in an L2 through the addition of new phonetic categories, or through the modification of old ones” (Flege, 1995, p. 233). Such notions are closely linked to a theory which has recently risen to prominence in the field of language learning and teaching, that of Cross-Linguistic Influence (CLI). CLI can be defined as “the influence of a person s knowledge of one language on that person’s knowledge or use of another language” (Jarvis & Pavlenko, 2008, p. 1). One of its main dimensions is linguistic transfer, which Jarvis and Pavlenko (2008) further divide into several subcategories, namely lexical, semantic, phonological, orthographic, syntactic, morphological, discursive, sociolinguistic and pragmatic. Regarding phonological transfer, it has long been accepted within CLI theory that the greater “the degree of difference between two phonological systems” (Rogerson, 2011, p. 20),1 the more difficulties a learner will have in L2 pronunciation. An example of two languages phonologically distant from the standard General British variety of English (GB) are Castilian Spanish and Galician, two of the official languages in Spain. Thus, it is commonly thought that the principal problems that Spanish students face in English pronunciation are due to these differences at a phonological level (Alcaraz & Moody, 1999; Estebas, 2012; Kenworthy, 1987; Palacios-Martínez, 2001). Consequently, over recent decades more and more scholarly attention is being paid to different aspects of the teaching and learning of English pronunciation in speakers of L1 Spanish. More specifically, we can distinguish two main types of studies: • Those that compare differences between the phonological systems of English and Spanish at a rather theoretical level, for instance, Alcaraz and Moody (1999), Avery and Erlich (1992), Coe (2001), Estebas (2012), Kenworthy (1987), PalaciosMartínez (2001), Parkinson (1983), Sánchez-Benedito (1994) and Walker (2010), to mention a few. • Studies in which practical experiments are conducted to identify specific difficulties that Spanish learners of English have with English pronunciation, or to test different approaches towards helping speakers of L1 Spanish improve their pronunciation of English. These include Aliaga (2007, 2009), Avello et al. (2012, 2013), Calvo-Benzies (2011), Carlisle (1998), Cebrian (2009), Cebrian and Carlet (2012, 2014), Cenoz and García-Lecumberri (1997, 1999), Cerviño and Mora (2009), Cutillas (2001), Gallardo-del-Puerto and Friedman (2013), Gallardo-del-Puerto et al. (2005), García-Landa (1999), Gómez-Lacabex and García-Lecumberri (2007), Gómez-Lacabex et al. (2005), Kivistö-de Souza and Mora (2014), Lara et al. (2014), Llanes et al. (2017), Llurda (1997), MartínezAdrián et al. (2013, pp. 51–83), Mompeán (2001), Mora and Valls-Ferrer (2012), 1 Phonological

distance.

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

23

Table 1 Some examples of studies on Spanish students’ problems with English pronunciation Vowels

Consonants

Intelligibility, accuracy, fluency and/or foreign accent

Effects of studying abroad

Intonation and/ or stress

Aliaga (2007, 2009), Calvo-Benzies (2011, 2013, 2014), Cebrian (2009), Cenoz and GarcíaLecumberri (1997, 1999), Cerviño and Mora (2009), Gómez-Lacabex et al. (2005), Gómez-Lacabex and GarcíaLecumberri (2007), Kivistö-de Souza and Mora (2014), Mompeán (2001), Safronova and Mora (2012)

Calvo-Benzies (2011, 2014), Carlisle (1998), Cebrian and Carlet (2012, 2014), Cutillas (2001), Gallardo-delPuerto and Friedman (2013), García-Landa (1999), Mora et al. (2014), Roothooft (2012)

Avello et al. (2012, 2013), Gallardo-delPuerto et al. (2005), Lara et al. (2014), Llanes et al. (2017), Llurda (1997), Mora and Valls-Ferrer (2012)

Avello et al. (2012, 2013), Lara et al. (2014), Llanes et al. (2017), Mora and Valls-Ferrer (2012)

RamírezVerdugo (2006)

Mora et al. (2014), Ramírez-Verdugo (2006), Roothooft (2012) and Safronova and Mora (2012), among many others. Most of the more theoretically-based studies provide explanations of the main differences between English and Spanish phonetics in order to warn Spanish learners of English as a Foreign Language (EFL) of such differences, and thus to try to reduce the number of mistakes they make when speaking in the FL. Moreover, Estebas (2012) and Palacios-Martínez (2001) go into more detail by outlining a series of recurrent problems that Spanish learners tend to encounter in English pronunciation due to the interference of their L1, and suggest some practical activities addressed to Spanish students to overcome such difficulties. The second group of studies can be divided into several groups. As Table 1 shows, we can find: • • • •

Studies which focus on students’ problems with different English vowels. Research into the acquisition of English consonants by Spanish learners. Studies on intelligibility, accuracy, fluency or foreign accent. Studies on the effect of studying abroad to improve one’s pronunciation.

24

Y. J. Calvo-Benzies

• Research on Spanish students’ problems with suprasegmental aspects like intonation and stress. Surprisingly, of all the studies mentioned above, only one focuses specifically on phonological transfer, namely, Martínez-Adrián et al. (2013, pp. 51–83). The authors were interested in analysing syntactic and phonetic mistakes made by third-year secondary education Spanish EFL learners (who had been in contact with English for seven years) that could be explained by transfer from their L1/L2 (Spanish/Basque) into their L3 (English). The research material used was the story Frog, where are you? (Mayer, 2003), which students had to narrate. It was found that more transfer mistakes were made at the phonetic than the syntactic level, the most frequent being “replacement of novel phonemes by L1 sounds, lack of aspiration in stop sounds, spirantisation of stop sounds, and closure of fricative sounds” (Martínez-Adrián et al., 2013, p. 11). Despite the usefulness of this study, it only considered phonetic transfer related to consonantal sounds, not vocalic ones. Furthermore, as can be inferred from what we have said thus far, most research in this area has tended to compare Spanish and English. In contrast, and to the best of my knowledge, hardly any research has been carried out on the problems that bilingual speakers of Spanish and Galician have with English pronunciation, with the notable exception of Estebas (2012) and Palacios-Martínez (2001). My own MA dissertation (Calvo-Benzies, 2011) included an experiment aimed at identifying and analysing the pronunciation mistakes made by bilingual speakers of Spanish and Galician in carrying out oral tasks. Some of the data used in this paper has been drawn from that study, and from a book chapter and journal article written subsequently (see Calvo-Benzies, 2013, 2014).2 However, in the present paper data taken from another research instrument have been added. Also, whereas the previous study looked only at the actual mistakes made, the analysis below will consider processes of phonetic and orthographic transfer. To summarise, it can be said that: (a) most available studies simply outline the main differences between Spanish and English at a phonological level, without adding any type of empirical analysis on mistakes most frequently made in terms of issues such as proficiency level and task type; (b) another group of studies has focused on analysing Spanish students’ problems, but with attention normally to particular segmental or suprasegmental aspects; (c) very few studies have compared the phonological system of Galician to that of English; and, (d) hardly any attention has been paid to phonological transfer in the spoken language of Spanish speakers, with the exception of Martínez-Adrián et al. (2013, pp. 51–83). Hence, the present study seeks to contribute to the field in several ways. First, it describes the phonological and orthographic interference of bilingual speakers of Spanish and Galician when speaking English. Second, it looks at transfer not only involving consonants (as in Martínez-Adrián et al., 2013, pp. 51–83) but also vowels and word stress. Finally, whereas the traditional data-gathering device for this type of research has been picture-stories, the present study, along with this approach, uses two further research instruments, namely a reading task and oral presentations. 2 This

present book chapter was published with the permission of Cambridge Scholars Publishing.

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

25

The specific research questions, then, are: 1. Do bilingual Spanish and Galician speakers make pronunciation mistakes due to interference from the phonological or orthographic system of their native languages? 2. Are there differences in the degree of phonological and orthographic transfer across the different tasks? 3. What are the main types of phonological and orthographic transfer found in the English production of bilingual Spanish and Galician speakers? 4. Will students with a lower level of proficiency in English make fewer mistakes due to orthographic and/or phonological interference than more advanced learners?

1.1 An Overview of the Orthographic and Phonological Systems of Spanish/Galician and General British Broadly speaking, the Spanish and Galician phonological systems differ considerably from that of GB. L1 bilingual speakers of Spanish and Galician have to accommodate quite a few new and similar sounds when they learn English, even though there are also some identical sounds in the phonological systems of Galician and English (similarities which do not occur in Spanish); it seems feasible that such identical sounds would not pose difficulties for native speakers of Galician when speaking English. Due to reasons of space, a detailed explanation of each of the vowels, diphthongs and consonants in Spanish, Galician and English cannot be provided. Also, it should be noted that attention will only be paid to differences at a segmental level. Walker (2010, p. 130) claims that one can always predict the pronunciation of a Spanish word simply by its spelling, whereas in the case of English there are far fewer direct correspondences between the phonological and orthographic systems. In other words, “English has what can be called an opaque orthography, in which phoneme-to-grapheme correspondences are inconsistent” whereas Spanish and Galician have “more transparent orthographies (…) in which phoneme-to-grapheme correspondences are more consentient” (Erdener & Burnham, 2005, p. 192). A direct consequence of this substantial difference is that lower-level Spanish students will “pronounce every letter, and this can lead to problems with intelligibility” (Walker, 2010, p. 130). Some examples illustrative of these major differences at the orthographic level in Spanish and English can be found in Table 2.3 For instance, the Spanish/Galician diphthong /ei/ can be represented by two spellings, whereas the corresponding English diphthong /eI/ can be spelled in six different ways. Regarding differences at the phonological level, perhaps the greatest difference is in the number of vowel sounds that each language has. Castilian Spanish has only five monophthong vowels /a, e, i, o, u/ whereas seven vowels are distinguished in 3 The

English examples were taken from Marks and Bowen (2012, pp. 32–35).

26

Y. J. Calvo-Benzies

Table 2 Spelling-sound relationships in Spanish/Galician and English for /ei, eI/ and / s/ Spanish/Galician phoneme

Spellings in Spanish and Galician

English phoneme

Spellings in English

/ei/

(reina, feira) (rey)

/eI/

(late) (train) (say) (great) (eight) (grey)

/s/

(sapo, detrás)

/s/

(say) (science) (cent)

Table 3 Vowels and consonants in General British, Castilian Spanish and Galician General British

Castilian Spanish

Galician

/æ, A , e, 2 , I, i , A, O , *, u , ђ, 2/

/a, e, i, o, u/

/a, e, i, o, u, 2, O/

Diphthongs

/AI, eI, OI, A*, ђ*, Iђ, eђ, *ђ/

/ai, au, ei, eu, oi, ia, i.e., io, ue, ua, ui, iu/

/ia, i.e., io, iu, ua, ue, ui, uo, ai, au, ei, eu, oi, ou/

Triphthongs

/AIђ, eIђ, OIђ, A*ђ, ђ*ђ/

/iai, iei, ioi, iau, uai, uei, uau/

/uai, uei, iai, iei/

Consonants

/p, b, m, f, v, ð, θ, t, d, s, z, n, l, r, Ѐ, Z, Ã, Ù, k, g, N, h/

/p, b, m, f, t, d, s, R, r, l, n, y, ¸l, ñ, k, x, g, θ, j/

/p, b, m, f, t, d, s, R, r, l, n, Ѐ, Ù, k, g, N, ñ, L/

Semivowels

/w, j/

——–

———

Monothong vowels

 

 

 

 

 

Galician /a, e, 2, i, o, O, u/; however, this figure increases to 12 for English /æ, 2, a , e, 2 , I, i , A, O , *, u , ђ/ (see Table 3). Another notable difference regarding vowels is duration. More specifically, Spanish and Galician do not make distinctions between short and long vowels, whereas GB does. For this reason, vowels like /2 / or / a / have been shown to be difficult for Spanish and Galician learners of English. Generally speaking, Spanish speakers typically pronounce the English sound with a similar vowel quality as the Spanish/Galician one. Estebas (2012, p. 13) explains that, regarding quality, English /i / should not be a problem for Spanish speakers “since English / i / and Spanish /i/ are almost identical in quality”; nevertheless, problems may arise when trying to produce the long English /i / in that duration is not a distinctive feature of the Spanish and Galician vowel systems. Finally, /2/ and / ђ/ are two vowel sounds that do not exist in the Romance languages under analysis here, and have been reported in the literature to pose many difficulties for Spanish learners of English. Quite a few similarities can be found between GB and Spanish/Galician diphthongs. For instance, the diphthongs /aI, eI, OI/ appear in the Castilian Spanish words  

 

 

 

 

 

 

 

 

 

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

27

paisaje ‘landscape’, reina ‘queen’, coincidir ‘coincide’ and in Galician caixa ‘box’, meiga ‘witch’ and oito ‘eight’. Generally, the only cross-linguistic differences that can be observed between Spanish/Galician and English diphthongs relate once more to quality. Thus, in / a*/ “the / */ quality of the final vowel is a bit more relaxed than the Spanish /u/” (Estebas, 2012, p. 35). Despite the aforementioned similarities, there are four diphthongs that exist neither in Spanish nor in Galician, these ending in either /*/ or /ђ/ , thus /ђ*, Iђ, eђ, *ђ/. Similarly, none of the five English triphthongs, /aIђ, eIђ, OIђ, a*ђ, ђ*ђ/, exist in Spanish or Galician, in that they all end with a schwa, a sound which, as already noted, does not form part of the phonological system of these two languages. Concerning consonants, the first difference that can be observed is the fact that Spanish and Galician only have the sounds /f/ and / θ/ whereas, GB distinguishes between a voiceless and a voiced sound for each of them, namely, /f, v/ and /ð, θ/, respectively. Similarly, GB distinguishes between the voiceless sound /s/ and the voiced consonant /z/. In contrast, only /s/ is a phoneme in Spanish and Galician; however, as Palacios-Martínez (2001, p. 23) points out: although /z/ does not exist as a phoneme in Spanish or Galician, it does exist as an allophone in Spanish words like mismo, asno and in Galician in forms such as trasno, sismo.

Consonants that do not exist in Spanish or in Galician include English /Z, Ã, j, w/. Similarly, Castilian Spanish does not have the fricative sound /Ѐ/ as found in English words like shoe, shower, or the /N/ sound in longer, watching. However, these are identical to the phonemes found in Galician words with ‘x’ (Xunta ‘The Autonomous government of Galicia’, Xosé ‘Jose, proper noun in Spanish’, xogo ‘game’) and ‘nh’ (algunha ‘some’, nunha ‘in one’) spellings. Differences are also found regarding English /h/, since this sound does not exist as a phoneme in Spanish or Galician; it is indeed represented in the spelling of some words, but is always a silent letter and remains unpronounced (as in hospital ‘hospital’, ahí ‘there’). Galician speakers may once again have an advantage here, since there is a phenomenon known as gheada in some parts of Galicia in which the phoneme /g/ in words like gato ‘cat’, barriga ‘tummy/stomach’ is pronounced as a Spanish but with aspiration. Hence, Galician students able to make such a realisation in their own language simply need to learn how to pronounce English /h/ without aspiration. Another difference between English and Castilian Spanish and/ or Galician is that the latter languages do not pronounce sounds like /p, t, k/with aspiration. English /ô/ is pronounced very differently in Spanish and Galician. More specifically, Spanish has two types of r-sounds, the tap /r/ and the trill /r/, respectively, but “none of the Spanish r-sounds corresponds to the English one” (Estebas, 2012, p. 92). Finally, there is a slight difference between English /b, d, g/ and Spanish/Galician /b, d, g/ in initial and final position; more specifically “in Spanish these sounds are voiced (produced with vocal fold vibration), in English they are devoiced (produced with no vibration of the vocal folds)” (Estebas, 2012, p. 56). In medial position these sounds are fully voiced when surrounded by vowels in English; in Spanish, however,

28 Table 4 Number of students taking part in this study, according to the educational level

Y. J. Calvo-Benzies

Non-university students Third year compulsory secondary education

First year noncompulsory secondary education

3

3

University students

Total

15

21

they are pronounced “with friction, that is, with a small noise caused by the air passing through a constriction somewhere in the mouth” (Estebas, 2012, p. 61). Consequently, when Spanish/Galician speakers pronounce these sounds between vowels in English words, they usually “pronounce them with friction causing a strong foreign accent effect” (Estebas, 2012, p. 61). In terms of word stress, generally speaking we can say that some cognates like English difficult and Spanish/Galician difícil are stressed on different syllables (in this case, the English word is stressed on the first syllable, and the Spanish word on the second); hence, as Palacios-Martínez points out (2001), Spanish/Galician speakers will tend to stress some polysyllabic English words on the wrong syllable by transferring the stress pattern from their native language.

2 Method As noted in Sect. 1, the main aim of this study is to identify and describe examples of phonological and orthographic interference in the oral production of bilingual speakers of Spanish and Galician when carrying out different types of spoken tasks in English.

2.1 Participants A total of 21 students from different educational levels took part in this study (see Table 4). Six of these were high school students studying English, three in their thirdyear of compulsory secondary education and three in the first-year of non-compulsory secondary education. The remaining 15 participants were university students in either their third or fifth-year of the old five-year degree in English Philology.4 The age of the students ranged from 14 to 26 years at the time of the study, and all were bilingual speakers of Spanish and Galician, studying English as a FL. Since the data were collected in different academic years, the subjects who took part in the picture-story and the reading tasks were different from those who partic4 A degree course which, due to the Bologna system, has now become a four-year degree called BA

in English Language and Literature.

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

29

Table 5 Number of students who completed each task, according to level of proficiency Third year compulsory secondary education

First year noncompulsory secondary education

Third year university degree

Fifth year university degree

Total

Task 1: Picture-story

3

3

3

3

12

Task 2: Reading

3

3

3

3

12

Task 3: Oral presentation

0

0

0

9

9

ipated in the oral presentation. Although this is clearly a disadvantage, the present study is a preliminary, descriptive analysis which aims to describe the main problems that native speakers of Spanish and Galician have with English pronunciation due to transfer at a phonological or orthographic level. Therefore, although some comparisons will be made where possible within the different groups and across the different tasks, further research with more comparable subjects is needed in order to establish truly empirical descriptions. Table 5 shows the number of students who completed each task, according to level of proficiency at the time the study.

2.2 Materials Three different research instruments were used to collect data: (a) a picture-story with a closed beginning and ending; (b) a reading task; and, (c) an oral presentation. These three activities were chosen because it was believed that, due to their heterogeneous nature, the results obtained in each could be complementary. More specifically, in the picture-story task speakers were not able to pre-plan what they wanted to say, and thus their production was spontaneous; furthermore, in this task participants had no written input, only some visual aids-the pictures themselves-which may have given them clues about which words to use but not about how to pronounce them. By contrast, in the reading task participants had all the visual input they needed in order to carry out the activity, that is, the exact written words which they were asked to read out. Finally, the oral presentation task can be classified as broadly pre-planned/nonspontaneous, in that students were given the opportunity to rehearse their speech several times before delivering it to their teacher and their classmates; nevertheless, it would be very difficult to memorise every word and thus the activity can also be considered as partially spontaneous, since students had to improvise at least some of their speech during the actual presentation. In addition, in this latter activity, it is common for speakers to have some written input, such as notes or a Powerpoint or Prezzi presentation, giving them some clues as to what to say. Figure 1 shows

30

Y. J. Calvo-Benzies

Fig. 1 Research instruments classified according to degree of spontaneity and amount of written input

Photo description

Reading

Less spontaneous

More spontaneous Reading

Oral presentation

WriƩen input

Oral presentaƟon

Photo descripƟon

No wriƩen input

a classification of the three research instruments used according to the degree of spontaneity and the amount of written input.

2.2.1

Picture-Story Task

Students were asked to make an oral description of a number of pictures that portrayed a man at several moments during a bad day. This task was taken from the course book New Generation 2 published by Granger and Beaumont (1987, p. 48).

2.2.2

Reading Task

The 160-word text Downloading Music was taken from the textbook Oxford Spotlight 3 (Davis & Falla, 2005, p. 100) for the students to read aloud. This text was chosen because it contained examples of most of the vocalic and consonantal sounds in GB.

2.2.3

Oral Presentation

Participants were required to give an oral presentation in small groups, which consisted of explaining how they would teach a particular grammatical, lexical, phonological or cultural aspect of English to a class of compulsory or non-compulsory secondary education students. The oral presentations of three pairs and one group of three learners will be analysed; Table 6 shows the main topic that each of these pairs/groups worked on.

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

31

2.3 Data Analysis A database was created to log the different pronunciation mistakes made by participants in each of the tasks. Each word was then analysed individually in order to identify those in which the mistake had been made due to the interference of Spanish and/or Galician. Following Jarvis and Pavlenko (2008), the different examples of language interference registered were then classified into two groups: • Orthographic transfer, that is, those cases in which a learner pronounced a word as in its orthographic form, following the transparent pattern of correspondences between phonology and orthography in Spanish and Galician (see Sect. 1.1); • Phonological transfer, when a mistake could be explained by comparing the phonological systems of Spanish and/or Galician to the GB one. Mistakes were classified by taking the GB model into account, since it continues to be the variety most commonly taught in EFL classes in Spain and most frequently found in the teaching materials used. On some occasions, examples of transfer identified here might not have been labelled as mistakes if another variety of English had been considered as the base of the study. For example, participants made many mistakes when pronouncing the spellings as /er, ar, or/. In the vast majority of cases, the previous set of spellings are pronounced with a schwa and without an /ô/ sound in GB; therefore, pronouncing the latter consonant within this study would be considered as an example of orthographic transfer.5

2.4 Results and Discussion The following section will be divided into two main parts. First, the general findings in each of the research instruments used (picture-stories, readings and oral presentations) will be treated separately. Following this, the orthographic and phonological

Table 6 Oral presentation topics chosen by students Topic chosen

Number of students working on this topic

Comparative and superlative forms of adjectives (grammatical)

2

Vocabulary inside and outside the classroom (lexical)

3

Basic differences in GB and General American (GA) pronunciation (phonetics and phonology)

2

Prepositions (grammatical)

2

5 The

pronunciation of final would not be considered as a mistake in some other standard varieties of English, for example, in General American (GA).

32

Y. J. Calvo-Benzies

interferences registered in each of them will be discussed. Finally, we will outline some comparisons that can be made across the different tasks and the different groups of participants.

2.4.1

Picture-Story Task

A total of 111 transfer mistakes in students’ pronunciation were registered in the picture-story task; as Table 7 shows, over three quarters of the problems identified entailed a transfer at the orthographic level, and the remaining 22.52% of mistakes involved phonological interference. Five different types of interferences at the phonological level were identified in this task. To begin with, on five occasions the velar / N/ at the end of the –ing verbal forms ringing, going and being was not pronounced. Broadly speaking, these mistakes can be explained by the fact that, although this velar consonant is an allophone of the nasal / n/ consonant in Castilian Spanish and is a phoneme in Galician (in words like nunha ‘in one’, dunha ‘of one’), in neither case can this sound appear in word-final position. On five different occasions, an epenthetic vowel was introduced before the initial consonant cluster , three times in the word stop and once in start; moreover, one example of an epenthetic vowel before the cluster was also registered in spent. This mistake can be explained by the fact that in Spanish (and Galician), “there are no words which start with an [s] followed by a consonant” (Estebas, 2012, p. 98). In addition, “one of the main differences between English and Spanish final consonant clusters is the number of consonants allowed at the end of a syllable” (Estebas, 2012, p. 103). More particularly, whilst in Spanish and Galician words we can find a maximum of two consonants in syllable final position, in English there can be up to four consonantal sounds; this could explain why some participants did not pronounce /t/ in crashed or /θ/ in sixth and seventh. The fourth type of phonological transfer identified in this first activity was the deletion of /h/. This sound is a deaf sound in Spanish and Galician (h muda ‘silent h’), not being pronounced at all in words like hospital ‘hospital’, búho ‘owl’, ahorros ‘savings’ (that is, not normally pronounced in initial, medial or final position), and this is most likely the reason why the /h/ sound was omitted by this subject.6 Finally, on 11 occasions different participants used an incorrect stress pattern in words like elevator or telephone. As shown in Table 8, most of the incorrectly-stressed words Table 7 Number of cases of phonological and orthographic interference identified in the picturestory task Type of mistake

Phonological

Orthographic

Total

25 (22.52%)

86 (77.48%)

111 (100%)

6 However, it is important to mention that h-deletion is considered grammatical in English, especially

in weak forms, and thus it would also be possible to consider this example as not a mistake.

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

33

Table 8 Types of phonological interference identified in the picture-story task Type of phonological interference

Number of mistakes found (total and percentage)

Items in which the mistakes were identified

Velar n

5 (20%)

2 (ringing) 2 (going) 1 (being)

Epenthetic vowel

5 (20%)

3 (stop) 1 (spent) 1 (start)

Consonant clusters

3 (12%)

1 (crashed) 1 (sixth) 1 (seventh)

h-deletion

1 (4%)

1 (has)

Wrong stress pattern

11 (44%)

5 (elevator) 2 (telephone) 1 (comfortably) 1 (continued) 1 (accident)

Total: 25 (100%)

recorded in this task correspond to cognates between the students’ native languages and English. Thus, the word accident is accidente in both Spanish and Galician. These words are distinguished in that the English item is based on three syllables whereas the Spanish and Galician words have four; moreover, regarding word stress, the correct way of realising the English form is by stressing the first syllable, / æk/ whereas the Spanish and Galician cognates are stressed on the third syllable / den/. This latter stress pattern, then, explains why one subject stressed the English word accident on the last syllable / dent/. Turning to transfer at the orthographic level, as can be inferred from Table 9, most of the orthographic interference identified in the picture-stories involved either pronouncing an English vowel as it appears in the orthographic form (i.e., following transparency patterns) or pronouncing the schwa in GB spellings as a full vowel, guided once again by the spelling, and also adding a Spanish /R/ sound. More specifically, Spanish /a/ was used on 19 occasions instead of the corresponding /A / (for example, in past, bath, bathroom, can’t or fast), /ђ/ (in machine, alone, physical) or /O / (in talking). A Spanish /e/ was used instead of schwa in broken, dozen and accident. Eight additional mistakes were recorded in which subjects pronounced an with the Castilian Spanish vowel /o/ instead of /2/ (in dozen), /ђ/ (in second, person) and /ђ*/ (in alone, telephone). Finally, Spanish /u/ was used on two occasions to pronounce tube (in Youtube) and continued. On the other hand, some examples in which the participants pronounced what should have been a /ђ/ with no pronunciation of the as Spanish /eR, oR, aR/were answer, shower, person, other, remember, more, forgot, forty, floor, morning, alarm and car. In other words, the aforementioned spellings were pronounced in a similar  

 

34

Y. J. Calvo-Benzies

Table 9 Types of orthographic interference identified in the picture-story task Type of orthographic interference

Number of mistakes found

Items in which the mistakes were identified

Type of orthographic interference

Number of mistakes found

Items in which the mistakes were identified

Spanish a

19 (22.09%)

1 (past) 6 (bath) 6 (bathroom) 1 (machine) 1 (fast) 1 (can´t) 1 (physical) 1 (talking) 1 (alone)

o+r

27 (31.39%)

1 (more) 3 (forgot) 5 (morning) 2 (forty) 3 (floor) 2 (fourth) 1 (for) 1 (important) 1 (officers) 1 (pouring) 1 (corner) 5 (elevator) 1 (comfortably)

e+r

15 (17.44%)

2 (answer) 2 (shower) 1 (another) 1(other) 1 (number) 1 (person) 2 (water) 1 (corner) 1 (remember) 1 (quarter) 1 (officers)

Spanish o

8 (9.3%)

3 (telephone) 2 (second) 1 (person) 1 (dozen) 1 (alone)

Spanish e

3 (3.48%)

1 (broken) 1 (dozen) 1 (accident)

Problems with /Ù/

1 (1.16%)

1 (situation)

a+r

9 (10.46%)

5 (alarm) 4 (car)

/Ѐ/ sound pronounced as / Ù/

1 (1.16%)

1 (machine)

Spanish u

2 (2.32%)

1 (Youtube) 1 (continued)

Problems with /Ã/

1 (1.16%)

1 (damaged)

Total: 86 (100%)

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

35

way to how these students would say words like saltar ‘to jump’, nacer ‘to be born’ and dolor ‘pain’ in their native languages. Apart from the previous two major types of phonological interference identified in this task, one example of situation pronounced as /sItu eIЀђn/, machine as /mђ Ùi n/ and damaged as /dæ maxed/ were also recorded. These mistakes once again represent examples in which subjects pronounced some English sounds as they appear in the orthographic form, more specifically, as /tu/, as /Ù/ and as Spanish /j/, respectively.  

2.4.2

Reading Task

Most cases of interference identified in the reading task involved transfer from the orthographic systems of students’ native language(s) onto their pronunciation of English (70.54%). An additional 29.45% correspond to transfer at the phonological level (Table 10). A total of 38 examples of phonological transfer were found in the reading task; these can be divided into five main types (see Table 11). Most frequent was the pronunciation of an incorrect word stress pattern (with 25 mistakes recorded). As in the previous task analysed, the English items that were stressed incorrectly in this activity have similar Spanish and/or Galician cognates which are stressed on a different syllable. More specifically, Spanish ilegal, legal and internet are all stressed on the final syllable whereas in industria ‘industry’, the prominent stress is on the second syllable. These different stress patterns clearly explain then why twelve, eight and three students respectively stressed illegal, legal and internet on the last syllable, and two participants stressed industry on . Four subjects failed to pronounce the final in the consonant cluster /ps/ in shops. This phenomenon may again be explained by the fact that Spanish and Galician syllables are typically closed by just one consonant, complex codas being very infrequent (e.g., Sanz). Similarly, five examples of inserting an epenthetic /e/vowel before the initial consonant cluster /st/ were also detected in this task (two in started, two in starting and one in store). The remaining four cases of phonological interference found were due to the non-pronunciation of /N/ in songs (by one participant) and the pronunciation of an English word as its similar Spanish counterpart, with per cent pronounced as /por sent/ by three subjects. As can be inferred from Table 12, 56 of the 91 cases of orthographic transfer identified in the reading task were related to the pronunciation of a Spanish monophthong vowel, following the spelling of the English word. To begin with, 18 mistakes were

Table 10 Number of cases of phonological and orthographic interference identified in the reading task Type of mistake

Phonological

Orthographic

Total

38 (29.45%)

91 (70.54%)

129 (100%)

36

Y. J. Calvo-Benzies

Table 11 Types of phonological interference identified in the reading task Type of phonological interference

Number of mistakes found

Items in which the mistakes were identified

Consonant clusters

4 (10.52%)

4 (shops)

Epenthetic vowel

5 (13.15%)

2 (started) 2 (starting) 1 (store)

/N/

1 (2.63%)

1 (songs)

Pronouncing /z/ for Wrong stress pattern

3 (7.89%) 25 (65.78%)

3 (per cent) 12 (illegal) 8 (legal) 3 (Internet) 2 (industry)

Total: 38 (100%)

recorded involving the pronunciation of Spanish /a/ in words spelled with ; more specifically, Spanish /a/ was pronounced instead of /O / (in always) and /ђ / (in legal, illegal). In addition, (a) Spanish /o/ was used instead of /ђ/ (in computer, action) on nine occasions; (b) /e/ was pronounced instead of /i / (in legal and illegal eight and seven times respectively), /I/ (twice in largest) and /2 / ; and, (c) Spanish /u/ was used eight times in the place of /ju / (in UK, computer) and on three occasions instead of /ђ/ (in industry). Moreover, /er/ was pronounced in other, however, computer and Internet; /ar/ in largest and starting; and, /or/ in forty and more. Two of the remaining problems involved pronouncing the consonants and as /g/ and /t/ instead of /Ã/ and /Ѐ/, respectively, in largest and action. These mistakes may be due to the fact that /Ã/ and /Ѐ/ do not exist in Spanish, and thus can be considered examples of new sounds (Flege, 1997) that students have to learn when acquiring English pronunciation. Nevertheless, as mentioned above in the case of /N/, /Ѐ/ is also part of the consonantal system of Galician and is used in words written with , as Xunta ‘The Autonomous government of Galicia’, Xabier ‘a male proper noun in Galician’ and roxo ‘red’; thus, it may once again seem surprising that bilingual speakers of Spanish and Galician have problems with this sound. However, since only one mistake with this sound was registered in this task, it should probably not be considered as the norm. Finally, one subject read against with the /aI/ diphthong rather than with /eI/. The latter sound exists in both Spanish and Galician but it is always represented by the spelling or (rey ‘king’, reina ‘queen’, peine ‘comb’) and never by . The Spanish/Galician diphthong /aI/, on the other hand, is always represented in spelling by , as in baile ‘dance’, aire ‘air’ and gaita ‘Galician bag-pipe musical instrument’.  

 

 

 

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

37

Table 12 Types of orthographic interference identified in the reading task Type of orthographic interference

Number of mistakes found

Items in which the mistakes were identified

Type of orthographic interference

Number of mistakes found

Items in which the mistakes were identified

Spanish a

18 (19.78%)

1 (always) 8 (legal) 9 (illegal)

e+r

21 (23.07%)

5 (other) 2 (however) 10 (computer) 4 (Internet)

Spanish o

9 (9.81%)

8 (computer) 1 (action)

a+r

2 (2.19%)

1 (largest) 1 (starting)

Spanish e

18 (19.78%)

8 (legal) 7 (illegal) 2 (largest) 1 (per cent)

o+r

6 (6.59%)

2 (forty) 4 (more)

Spanish u

11 (12.08%)

2 (U.K) 6 (computer) 3 (industry)

a+i

1 (1.09%)

1 (against)

Pronunciation of /Ã/ as / g/

4 (4.39%)

4 (largest)

Pronunciation 1 (1.09%) of t + i instead of / Ù/

1 (action)

Total: 91 (100%) Table 13 Number of cases of phonological and orthographic interference identified in the oral presentations Type of mistake

2.4.3

Phonological

Orthographic

Total

10 (11.49%)

77 (88.51%)

87 (100%)

Oral Presentations

Over 85% of the mistakes made by the undergraduates in their oral presentations entailed transfer at the orthographic level whereas very few phonological interferences were registered in this task. Only two types of phonological interference were registered in this task. On the one hand, five examples of the insertion of an epenthetic vowel in initial position, two before the consonant clusters (in skills), two in front of (in students and start) and one before in specify. On the other hand, three participants stressed the word exercise on the third syllable, as is the case in its Spanish and Galician cognates, ejercicio and exercicio; similarly, two subjects stressed photocopies on the third syllable, again reflecting the stress pattern of its cognate fotocopias in the students’ native languages (Tables 13 and 14).

38

Y. J. Calvo-Benzies

Table 14 Types of phonological interference identified in the oral presentations Type of phonological interference

Number of mistakes found

Items in which the mistakes were identified

Epenthetic vowel

5 (50%)

2 (skills) 1 (specify) 1 (students) 1 (start)

Wrong stress pattern

5 (50%)

3 (exercise(s) 2 (photocopies)

Total: 10 (100%)

In this task only two types of orthographic transfer were identified. These coincide with the two most-productive kinds of phonological interference recorded in the previous two activities, that is: (a) pronouncing a monophthong vowel as it is spelled following the Spanish and Galician systems of transparency between phonology and orthography and, (b) pronouncing the combinations with a full vowel plus an /R/ sound (see Table 15). In the former type of transfer, students pronounced words such as word, oldest and motivate with /o/ instead of /2 /, /ђ*/ and /ђ/; was pronounced as /a/ instead of /ђ/, /a /, /I/ and /eI/ in items like superlative, example, garage or washbasin. In addition, the letter was pronounced as /e/ in oldest, different, toilet and students rather than with /I/ and /ђ/, and was pronounced as /i/ in dining room, furniture and washbasin. On the other hand, was pronounced /eR/ in words such as interaction, order, number and exercise, /oR/ in more, porch and important, and /aR/ in garden, partner and standard.  

 

2.4.4

Results Across Tasks and Groups of Participants

As Table 16 shows, the four most-productive types of phonological interference registered in the data across all three research materials were: (a) using an incorrect stress pattern; more specifically, reflecting the stress pattern of the corresponding cognate in Spanish and/ or Galician; (b) the introduction of an epenthetic vowel, always in word-initial position; (c) the pronunciation of /N/ as /n/, especially in –ing verbal forms; and, (d) problems pronouncing consonant clusters, mainly in final word position. Regarding orthographic interference, the majority of mistakes identified for each of the three tasks can be divided into two main groups: (a) the pronunciation of English spellings with the corresponding full Spanish vowel plus an /R/sound; and, (b) pronouncing a vowel (not followed by a rhotic consonant) with a full Spanish vowel guided by the spelling of the word. The final results obtained in this study indicate that the former problem accounts for nearly 45% of the total mistakes registered, whereas 52.74% of all cases of orthographic transfer involved the latter problem (see Table 17).

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

39

Table 15 Types of orthographic interference identified in the oral presentations Type of orthographic interference

Number of mistakes found

Items in which the mistakes were identified

Type of orthographic interference

Number of mistakes found

Items in which the mistakes were identified

e+r

16 (20.77%)

2 (superlative) 1 (other) 1 (answer) 1 (partner) 2 (under) 1 (remember) 1 (winner) 1 (interaction) 1 (teacher) 1 (order) 1 (interpretations) 1 (number) 2 (exercise(s)

Spanish o

20 (25.97%)

1 (word) 1 (second) 1 (oldest) 3 (vocabulary) 1 (translation) 1 (questions) 1 (complete) 1 (definitions) 1 (computer) 2 (preposition) 3 (photocopy/ies) 1 (instructions) 1 (motivate) 1 (position) 1 (directions)

o+r

7 (9.09%)

2 (more) 1 (porch) 1 (important) 1 (reinforce) 1 (according) 1 (bored)

Spanish a

18 (23.37%)

3 (superlative) 3 (comparative) 2 (example) 1 (typical) 1 (garage) 1 (explain) 2 (bathroom) 2 (bath) 1 (washbasin) 1 (ask) 1 (American)

a+r

8 (10.38%)

1 (irregular) 1 (garden) 3 (parts) 1 (car) 1 (partner) 1 (standard)

Spanish e

4 (5.19%)

1 (oldest) 1 (toilet) 1 (different) 1 (students)

Spanish u

1 (1.29%)

1 (furniture)

Spanish i

3 (3.89%)

1 (dining room) 1 (washbasin) 1 (furniture)

Total: 77 (100%)

Turning now to some comparisons between the number and types of transfer registered in the picture-story and the reading tasks, the first notable finding here is that the total number of cases of interference in both activities is quite homogeneous, 111 and 129, respectively. This may seem surprising since the picture-story and the reading activities are different in terms of spontaneity and written input; whilst the former is considered a spontaneous activity, the latter is not. Furthermore, in the reading task participants were explicitly given the words they had to pronounce in

40 Table 16 Types of phonological interference identified across the three research instruments

Table 17 Types of orthographic interference identified across the three research instruments

Y. J. Calvo-Benzies

Type of phonological interference

Number of mistakes found

Epenthetic vowel

15 (20.54%)

/N/

6 (8.21%)

Clusters

7 (9.58%)

h- deletion

1 (1.36%)

/z/vs. /s/

3 (4.1%)

Stress pattern

41 (56.16%)

Total

73 (100%)

Type of orthographic interference

Number of mistakes found

e+r

52 (20.47%)

Spanish a

55 (21.65%)

Spanish o

37 (14.56%)

o+r

40 (15.74%)

Spanish e

25 (9.84%)

Spanish u

14 (5.51%)

a+r

19 (7.48%)

Pronunciation of /Ã/ as / g/

5 (1.96%)

Spanish i

3 (1.18%)

Problems with /Ù/

2 (0.78%)

/Ѐ/ sound pronounced as /Ù/

1 (0.39%)

a+i

1 (0.39%) 254 (100%)

written form, whereas in the picture-story task only visual input was given. This finding, then, is interesting because it may indicate that on many occasions Spanish EFL learners are continuously thinking of the orthographic form of a word, and not only when they are provided with the written form. Broadly speaking, since the participants of this study belong to different proficiency levels it would seem feasible that lower-level students, in this case those in compulsory secondary education, might make more mistakes than those studying a university degree in English Philology. Also, the expected pattern regarding orthographic and phonological transfer would be that less proficient students are more influenced by their native languages when speaking English. This is confirmed, since, as shown in Table 18, the highest percentage of all orthographic and phonological mistakes recorded in the data in the picture-story and the reading tasks were made by third year compulsory secondary education students.

4 6 9 25

Third year Philology

Fifth year Philology

Total

26

86

26

17

17

111 (46.25%)

6

First year non-compulsory secondary education

91

8

15

34

34

Ortho.

129 (53.75%)

38

9

6

8

15

Phon.

Third year compulsory secondary education

Readings

Phon.

Ortho.

Picture-story

Table 18 Total number of cases of interference per group in the picture-story and reading activities

240 (100%)

52 (21.66%)

44 (18.33%)

63 (26.25%)

81 (33.75%)

Total

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* … 41

42

Y. J. Calvo-Benzies

Table 19 Number of words pronounced by each student in the picture-story task Third year compulsory secondary education

First year noncompulsory secondary education

Third year Philology

Fifth year Philology

S1

S2

S3

S4

S5

S6

S7

S8

S9

S10

S11

S12

Number of words

134

104

150

162

170

115

191

232

186

238

217

220

Average across groups

129.33

149

203

225

In general terms, then, these results indicate that the two lower proficiency groups of students (third year compulsory secondary education and first year noncompulsory secondary education) tend to make more mistakes than the two groups of university students throughout the two different tasks. Nevertheless, there is one exception, and that is when we compare the number of orthographic and phonological transfer made by participants in the third year of compulsory secondary education with those made by the fifth year English Philology students in the picture-story task. As shown in Table 18, six cases of phonological transfer and 26 of orthographic interference were registered in the data collected from the lowest-level students, and these figures are quite similar to the data for the final-year university students, who registered nine cases of phonological transfer and 26 orthographic ones. This may be explained by the fact that, as would seem natural in the language learning process, the higher the level of English proficiency a student has in English, the more words they will be able to use when facing a picture-story task, given that they will have more vocabulary and more complex grammatical structures to draw on in improvising their answers. As can be inferred from Table 19, this seems to be the case in the data collected for the present study, since the average number of words pronounced in the pictures stories by the students in the third year of compulsory secondary education was 129.33, whereas this figure rose to 225 in the case of the data collected for fifth-year university students. Finally in this section, we turn briefly to the number of mistakes made for each sub-type of transfer identified across the data, collected from the participants of lower and higher levels of proficiency in the picture-story and reading activities. For ease of presentation, the highest amount(s) of mistakes registered for each sub-type of phonological and orthographic transfer appear in bold font in Tables 20, 21, 22 and 23. In broad terms, the data in these tables complements, verifies and illustrates some of the previous findings very clearly. First, in the picture-story task the highest number of mistakes made by university students involved the introduction of an initial epenthetic vowel and incorrect stress patterns; as noted above, this can be explained by the fact that these students produced longer descriptions in this task and thus had higher chances of incorrectly pronouncing more English words. On the other hand, the results which can be extracted from the orthographic transfer registered in this descriptive task are somewhat varied, since students of a lower level made more

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

43

Table 20 Types of phonological interference identified across the different groups of students in the picture-story task Type of phonological interference

Third year compulsory secondary education

First year noncompulsory secondary education

Third year Philology

Fifth year Philology

Totals

Epenthetic vowel

0

2

0

3

5 (19.23%)

Velar n

1

0

3

1

5 (19.23%)

Problems with consonant clusters

1

1

1

0

3 (11.53%)

Omission of h sound

0

0

1

0

1 (3.84%)

/z/vs. /s/

0

1

0

0

1 (3.84%)

Incorrect stress pattern

3

1

1

6

11 (42.3%) 26 (100%)

Table 21 Types of orthographic interferences identified across the different groups of students in the picture-story task Type of orthographic interference

Third year compulsory secondary education

First year noncompulsory secondary education

Third year Philology

Fifth year Philology

Totals

Spanish a

4

3

4

7

18 (21.17%)

e+r

4

5

5

2

16 (18.82%)

Spanish e

1

0

0

2

3 (3.53%)

a+r

3

1

2

3

9 (10.5%)

Spanish u

0

0

0

2

2 (2.36%)

o+r

10

7

5

5

27 (31.76%)

Spanish o

3

1

1

3

8 (9.41%)

Problems with /Ù/

0

0

0

1

1 (1.17%)

Problems with /Ã/

0

0

1

0

1 (1.17%) 85 (100%)

44

Y. J. Calvo-Benzies

Table 22 Types of phonological interference identified across the different groups of students in the reading task Type of phonological interference

Third year compulsory secondary education

First year noncompulsory secondary education

Third year Philology

Fifth year Philology

Totals

Epenthetic vowel

3

0

1

1

5 (20%)

Velar n

0

0

0

1

1 (4%)

Problems with consonant clusters

2

1

1

0

4 (16%)

/z/vs. /s/

2

1

0

0

3 (12%)

Incorrect stress pattern

5

4

2

1

11 (44%) 25 (100%)

Table 23 Types of orthographic interference identified across the different groups of students in the reading task Type of orthographic interference

Third year compulsory secondary education

First year noncompulsory secondary education

Third year Philology

Fifth year Philology

Totals

Spanish a

7

6

3

2

18 (19.78%)

Spanish o

2

3

2

2

9 (9.89%)

Spanish e

6

8

2

3

19 (20.87%)

Spanish u

3

5

2

1

11 (12.08%)

Pronunciation of /Ã/ as/ g/

2

0

1

1

4 (4.39%)

e+r

6

7

3

4

20 (21.97%)

a+r

0

0

1

1

2 (2.19%)

o+r

2

3

1

0

6 (6.59%)

a+i

0

1

0

0

1 (1.09%)

Pronunciation of t + i instead of/ Ù/

0

1

0

0

1 (1.09%)

91 (100%)

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

45

mistakes of the type as /oR/, whereas the older and more proficient students made more mistakes by pronouncing a Spanish /a/ sound for spellings. In the reading task, the number of mistakes made by the high school students was greater in almost all the sub-types of both phonological and orthographic interference identified in this task. For example, nine mistakes were registered of incorrect stress patterns, three of introducing an initial epenthetic vowel, 13 of pronouncing with Spanish /es/ plus a rhotic consonant, and 14 of pronouncing a Spanish /e/ for in the data from the two pre-university groups; these figures are reduced to three, two, seven and five, respectively, in the data from university students. In contrast to the previous activity, all students here, regardless of their level of proficiency, had to pronounce exactly the same number of words since the reading text was the same; consequently, the results for this task follow the expected pattern, one in which the lower proficiency learners are more frequently influenced by their L1s than those with a higher level of FL proficiency, in terms of both phonology and orthography.

3 Conclusions and Implications for the Teaching of EFL in Spain Overall, the findings indicate that the phonological and orthographic systems of Spanish and Galician have a strong influence on the way bilingual speakers of these languages pronounce different words in English. A substantial number of examples of both phonological and orthographic transfer were found in the three research instruments used. As seen in Sect. 2.4, orthographic transfer was the most productive type of interference in the three tasks, with 86 mistakes registered in the picture-story task, 91 in the reading task, and 77 in the oral presentations. The fact that a similar amount of orthographic interference s was found in the first two tasks may suggest that bilingual speakers of Spanish and Galician constantly have the orthographic form of English words in mind, no matter whether they are performing a reading task or if their production is based on no written input. For instance, it could be said that the pronunciation by two university students of photocopy as /foto copi/ indicates that they were thinking about the orthographic form when they pronounced the word aloud and, following the transparent system of orthography and pronunciation in Spanish and Galician, decided to pronounce the word with several Spanish /o/ sounds, this guided by the spelling . Although more research is needed, the fact that similar orthographic and phonological types of transfer were registered in the data collected for each research instrument generally indicates that some sounds pose more problems for bilingual speakers of Spanish and Galician than others. In broad terms, most books on the teaching of pronunciation to Spanish EFL learners include tips and/ or activities to practise the whole inventory of English consonants and vowels. Similarly, EFL textbooks used in Spain at different educational levels tend to include the same segmental and suprasegmental features year after year (see Calvo-Benzies, 2016); according to the present

46

Y. J. Calvo-Benzies

findings, though, it might be the case that in the Spanish educational system more attention needs to be paid to certain sounds rather than others. So, although English /b/ and /v/ are both represented in Spanish and Galician as /b/, no mistakes were registered involving the pronunciation of what should have been a /v/ sound as /b/ in our data. Likewise, it has been pointed out that Spanish learners of English have serious problems with the consonant /Ã/ since it does not exist in Spanish; nevertheless, in this study hardly any examples of either phonological or orthographic transfer were registered for this sound. By contrast, the findings here suggest that special attention should be paid to /N/ and consonant clusters (in both initial and word-final position) in EFL classes in Spain. Moreover, the teaching of stress patterns in cognates like Internet, legal, illegal, million, exercises and photocopy might also be a useful area to focus on. In addition to the previous phonological aspects mentioned, the findings here indicate that bilingual students of Spanish and Galician very frequently rely on the way vowels appear in the written form of an English word; in fact, the vast majority of cases of orthographic transfer registered in the data collected for each research instrument entailed either pronouncing a single vowel as it is spelled, as would be done for Spanish and Galician words, or pronouncing the combinations with a full Spanish/Galician vowel plus a Spanish /r/ sound. Such findings reveal that bilingual speakers of Spanish and Galician: (a) have serious problems with schwa; and, (b) feel more comfortable pronouncing in every position, especially when it is preceded by the vowels and . Examples of these types of orthographic transfer were found across all the levels of proficiency studied, and in this sense Walker’s (2010) observation that it will be younger Spanish EFL learners who tend to read every single letter of an English word cannot be taken for granted, in that the present findings indicate that even advanced students of English–namely, undergraduate students in a BA in English Language and Literature–also resort to this technique of being guided by the orthographic transparency patterns of their L1s (see Sect. 1.1). As would seem natural in a study with these characteristics, a higher rate of both orthographic and phonological interference were registered in the data collected for the lower-level students in both tasks; however, as discussed in Sect. 2.4.4, an exception to this general rule was found in the data analysed for the picture-stories. More specifically, the number of orthographic interferences registered in the picturestories made by both the younger and older students was quite similar; however, this can be explained by the fact that the undergraduates made longer descriptions than those in secondary and post-secondary education and therefore had more chances of making pronunciation mistakes (see Table 19). Although more research is needed, and at this stage we cannot affirm that advanced Spanish EFL learners continue to make the same types of mistakes that younger ones do when no written form is available to them, we can infer from the general results that EFL students in Spain: • Are not receiving enough opportunities to practise English pronunciation; for instance, many mistakes were registered with words that students should be famil-

/ðђ musIk In dustrI jas es tarted teikin le gal ak ЀAn/* …

47

iar with and which they probably use frequently when communicating in English, such as computer, bathroom, morning, alarm, elevator, shops, car, Internet, photo, photocopy, corner, other, another, answer, teacher, exercise and water; it seems that insufficient attention is being paid to the pronunciation of these items, among others; and, • They are not familiar enough with certain differences and similarities between the orthographic and phonological systems of Spanish and English. Another interesting example within the data analysed concerns the word corner, which was pronounced by some participants as / korner/. The Castilian Spanish and Galician languages borrowed this word from English and adapted it to a more natural Spanish-like pronunciation / korner/. It seems that students sometimes forget that a word they have in their language was taken from English and have internalised the way it is pronounced in their L1s, assuming that it is pronounced in the same way in the original language, in this case English. In sum, this study has sought to contribute to the field of CLI in the English pronunciation of bilingual speakers of Spanish and Galician. Future studies could be conducted to complement and better explain the results obtained in this paper. First, it would be interesting to assess the extent to which some of the mistakes registered in this study are intelligible or not for native and other non-native speakers of English; that is, to test whether other speakers of English would understand a word like per cent if it is pronounced /por sent/. Second, it would be interesting: (a) to gather data from pre-university students giving an oral presentation; and, (b) to achieve greater control of the target words in the data analysis. In this study, the picture-story and especially the oral presentation were activities in which students could spontaneously decide what they wanted to say and how to say it, and thus the majority of words pronounced by one student in an oral presentation might not have been used by others in the same activity. Hence, it would be interesting to devise a means of assuring that all students have to pronounce at least some of the same words, through giving an oral presentation on the same topic, for example. Furthermore, the fact that not many mistakes were registered with some sounds, such as /Z, Ã, j, 2/, may indicate that bilingual students of Spanish and Galician either do not have problems with certain English consonants and vowels, and hence that attention should be focused elsewhere, on sounds which they do have problems with, or that there were not many words containing these sounds in the data analysed and for this reason no mistakes were found. So, again depending on the target words, an activity might be devised that obliges students to pronounce words containing the specific sound/s under analysis.

CLI in Lexical Accessibility María del Pilar Agustín-Llach

Abstract The present paper aims to explore the role of mother tongue (L1) influence while completing a lexical availability task. To learn more about how learners’ lexical knowledge is structured and accessed, and to look into their mental lexicon, a lexical availability task has been used (cf. Ávila-Muñoz & Sanchez-Saez, 2014). The assumed simultaneous activation of the native and target language in lexical accessibility in a lexical availability task might result in the L1 or any other previous linguistic knowledge playing a determining role (cf. Jarvis, 2009). In this sense, to look into the instances of lexical transfer can, therefore, be enlightening in order to learn more about the interface between lexical knowledge, lexical use, and lexical access. Accordingly, how L1 knowledge or knowledge of any other known language (Ln) influences second language (L2) lexical production in a lexical availability task of a group of Spanish pre-university students has been the focus of this study. Another point of interest is comparing how previous linguistic knowledge in a lexical availability task manifests in Spanish monolingual English as a Foreign Language (EFL) learners and learners who can speak other languages as L1, in addition to Spanish (bilingual Spanish-Ln EFL learners). Finally, some implications for the classroom derived from research-related results are discussed. Specifically, explicit lexical instruction and cross-linguistic lexical comparisons are called for with a focus on cognate use, and cognate awareness rising. Fostering and training in communication and compensatory strategy use can also help learners deal with vocabulary problems and lexical gaps. Keywords Cross-linguistic influence · Lexical availability task · L3 lexical learning · Lexical storage and access

M. del P. Agustín-Llach (B) Universidad de La Rioja, C/San José de Calasanz 33, 26004 Logroño, La Rioja, Spain e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_3

49

50

M. del P. Agustín-Llach

1 Introduction Cross-linguistic influence (CLI) at the lexical level is a widely-studied phenomenon (e.g., Arabski, 2006; Jarvis, 2009; Ringbom, 2006) which allows to draw conclusions on how learners set to the task of learning and using the vocabulary of a new language, and on the role of the first language (L1) or any other language (Ln), in that process. CLI studies tend to obtain data from learners’ written compositions or oral discourses, but using other types of tasks is not frequent. The present paper examines how CLI manifests in a lexical availability task, since it is believed that this will allow to delve deeper into how the mental lexicons of the L1 and second language (L2) are organized and what the mechanisms are that keep latent or activated the languages students possess. Accordingly, a revision of the main issues concerning CLI at the lexical level will start the literature review section followed by the exploration of the notion of lexical availability and discussion of why it is important for CLI studies, and finally, a brief account of the benefits of bilingualism in lexical learning will close this section. The account of the design and methodology of the study will follow, with presentation and discussion of results coming later and a conclusion closing the paper.

2 Literature Review 2.1 Cross-Lexical Influence CLI refers to the influence of a (previously known) language on another language (being learned), and has traditionally been given the name of transfer, and specifically, lexical transfer when the influence occurs at the lexical level. The notion of CLI takes into account influences, which go from the L1 into the L2, from the L2 into the third language (L3) or from the L3/Ln into the L2 or L1. This implies one-step forward into considering relationships of influence, which were not initially included in the notion of transfer. The study of CLI at the lexical level or lexical transfer, for short, has been going on for a long time. Already in the nineteenth century researchers were interested in the phenomenon of loanwords or borrowings (see Odlin, 1989 for a review of these). Arabski (2006), Ecke (2015), Jarvis and Pavlenko (2008) and Singleton (2016) provide reviews of issues related to CLI in the lexicon of L2 and L3 learners. The phenomenon of cross-lexical interactions can be examined from a dichotomous or two-sided perspective. First, the distinction between the resulting effects of the influence is taken into account. Positive cross-lexical influence can be based on the learners’ recognition of cognates. Cognates are words, which share formal and semantic specifications in the native or source language and in the L2 (cf. Odlin, 1989; Otwinowska, 2016). Cognates can be traced back to two origins. First, they might derive from common linguistic ancestors, for instance, words of Greek and

CLI in Lexical Accessibility

51

Latin origin, which abound in the lexical systems of languages at stake. Second, cognates might be, for instance, international words, words that were coined to describe or refer to new inventions or situations within a globalized world. Words such as telephone, or internet, which are common to most languages across the globe, belong to this group. Whatever their origin, the languages at stake in the present study, English and Spanish, share a great number of cognates, such as for instance, educación—education, radio—radio, oxígeno—oxygen. Cognates as manifestations of positive cross-lexical influence are believed to have a facilitating effect in lexical learning in the foreign language (FL) with the corresponding contribution to increase word knowledge (cf. Otwinowska, 2016). The presence of cognates implies that there is less to learn in terms of both form and meaning (and concept). Linguistic resources are thus maximized; with one single lexical entry, learners have knowledge of two languages. Accordingly, communication is made easier, with less cognitive load, and without modification or “retuning” of the mental lexicon (Jarvis, 2009; Otwinowska, 2016). Finally, the use of cognates when learning and using the FL can serve as a communication strategy to compensate for lack of lexical knowledge. Learners recognize and use formal similarity between new and already known words, as in the case of cognates, to effectively learn new words. Once they detect this formal similarity, they will assume by default that syntactic and semantic attributes (lemma/meaning) are the same and connect the new word form with the syntactic and semantic features of the L1 or Ln equivalent (see Hall, 2002; Hall et al., 2009). This will often be an appropriate connection. However, sometimes it can also lead to errors where there is a mismatch between equivalents in syntactic frames and meaning (as in the case of false friends or false cognates). This takes the discussion to negative influence. Cross-lexical influence can also have a negative effect when it derives in a lexical error and has the potential of interrupting or impeding communication, damaging the image of the speaker/learner and/or irritating the hearer, reader or interlocutor. However, and despite these undesirable effects, this so-called negative CLI can have some positive implications, since it helps scaffolding and thus learning; it compensates for lack of lexical knowledge, and can serve as a communication strategy. In fact, good language learners use this type of strategies, that is, good language learners transfer among their languages. Jarvis (2000) still includes a third possibility to complete the positive-negative dichotomy, which he termed lexical reference or lexical choice. This type of transfer results in evidence which does not derive in a lexical error, but in e.g. register, pragmatics, or style problems. In other words, it refers to a lexical choice which would not be frequent or preferred by a native speaker, because it is a non-orthodox or awkward selection. Finally, Ringbom (2006) took over Schachter’s (1974) idea and referred to still a fourth possible manifestation of cross-lexical transfer from the perspective of the resulting evidence, which is avoidance. Some words might be avoided because learners may not feel at ease with them. However, avoidance can only be identified through prior contrastive analysis, and not through learners’ productions.

52

M. del P. Agustín-Llach

From a procedural perspective, cross-lexical interactions might manifest in formal (lexemic, cf. Jarvis, 2009), semantic (lemmatic, cf. Jarvis, 2009) and conceptual transfer. Formal lexical transfer consists in the use of L1 forms when producing in the L2 (cf. de Angelis & Selinker, 2001; Ringbom, 2001, 2006). Different types of formal cross-lexical influence have been identified, for example borrowings or code-switching, when an L1 word is inserted into L2 syntax; phonetic rendering, sometimes also misspelling, when the L2 lexical item is written as it is pronounced by the learner; or lexical invention or word coinage, when it has an L1 base. Negative semantic cross-lexical interactions are generally derived from the transfer of semantic or meaning patterns of the L1 into L2 lexical items (Ringbom, 2001, p. 60). Literal translations or calques, semantic extensions, where the semantic features of the L1 are transferred into the new L2 lexical item, or unsuccessful word choice with source language base are different categories of semantic cross-lexical interactions. Finally, interference at the level of concepts is another type of cross-lexical interaction. Current research has given little attention to examining cross-lexical influence on lexical availability. In the next section, accounts of lexical availability and specifically on how lexical transfer influences lexical availability will be given.

2.2 Lexical Availability Lexical availability tasks intend to identify the words that spring to the student’s mind in response to different prompts or stimuli. The lexical availability tasks reflect the participants’ spontaneous vocabulary production. The lexical availability task asks learners to establish word relationships not only with the stimulus word or center of interest but also with the preceding words (words having been produced earlier in the task), also called chain associations. For a time lapse of two minutes per prompt, learners have to write as many words that come to their minds in relation to that prompt without any further restriction concerning formal or semantic relations. Studies on CLI and cross-lexical interactions in writing use composition tasks to gather data, i.e. open and communicative tasks. However, a lexical availability task can be interesting, due to its timed nature, to examine cross-lexical interferences and reveal more about cross-lexical connections and the influence of L1/Ln on lexical learning processes. Lexical availability of learners allows to look into their mental lexicon and to learn more about how their lexical knowledge is structured and accessed (cf. Ávila-Muñoz & Sanchez-Saez, 2014). In this sense, the L1 or any other previous linguistic knowledge is susceptible to play a role in lexical use through the simultaneous activation of the native and target language (cf. Jarvis, 2009), however there also might be other factors influencing lexical knowledge and use. To look into the instances of lexical transfer can, therefore, be enlightening to learn more about the interface between lexical knowledge, lexical use, and lexical access. Actually, one application of lexical availability is interlanguage analysis, processes of lexical generation, and transfer processes and interference identification (e.g., Paredes García, 2012; Tabernero, 2008).

CLI in Lexical Accessibility

53

Few studies have specifically explored the presence of CLI in this task, but some studies have examined the role of different L1s in lexical availability in Spanish FL to conclude that it plays no significant role in the words learners have available (Carcedo González, 2000). In a more recent study, Martínez-Adrián and Gallardodel-Puerto (2017) analyzed the lexical availability in EFL of Spanish and Chinese learners. They found differences among the groups of learners with Spanish learners showing higher levels of lexical availability and of spelling accuracy than Chinese learners. They allude to typological proximity to explain their results. Studies on word retrieval problems and tip-of-the-tongue (TOT) phenomenon show that the strength of CLI and its source will differ depending on the learners’ language proficiency and use patterns. For instance, Ecke (2008) investigated the word associations produced during TOT states in Spanish speakers with 3 different levels of English. Findings showed that CLI from English on L1 word search was stronger when the L2 proficiency was higher. In a similar vein, Ecke and Hall (2013) conducted a longitudinal study of word associations in naturally occurring TOT states of multilinguals. They found that the L1 was becoming less vulnerable to lexical retrieval failure, more often the source of CLI in other TOT states, and more resistant to CLI than other (more frequently used) languages. Among the studies on lexical availability that include CLI in their research agenda, Tabernero (2008) tested Basque-dominant bilingual informants’ lexical availability in Spanish and looked for instances of lexical transfer, between the two languages. She found that phonetic rendering or phonetic spellings were a very frequent category in her Spanish-Basque bilingual informants. Similarly, she could also attest many instances of code switching when the center of interest referred to communicative contexts where speakers used the other language, Basque in this case. The incorporation in the responses of adapted or literal cultural words in Basque was frequent, as well, especially in the domains where learners felt the culture originated in the Basque Country or Navarre. In what follows, exploration will be made of the interface between bilingualism, lexical availability and cross-lexical influence.

2.3 Bilinguals, Cross-Lexical Influence and Lexical Availability Studies on bilingualism and bilingual learners have been a frequent and vigorous focus of research since long. However, studies addressing how bilingual learners learn a new language and how this process compares to that of monolingual learners learning a FL are not so common to our knowledge. Previous studies (e.g., GabrysBarker, 2006; Kroll, Bogulski, & McClain, 2012; Otwinowska, 2016) have concluded that bilingual learners learning a FL have a higher repertoire of learning strategies, an increased metalinguistic awareness, better memory performance, and are more verbally creative. Additionally, and this is relevant for our current purposes, bilingual learners are able to make conscious cross-linguistic comparisons and thus are

54

M. del P. Agustín-Llach

believed to benefit more largely from cognateness effects (cf. Otwinowska, 2016). Similarly, Kroll et al. (2012) believe that bilingualism has a series of benefits in cognitive tasks, and that bilinguals can better control the inhibition of one language when using the other. Accordingly, one could assume that CLI would be less or of a different kind in bilingual and monolingual learners. And in relation to this, Kroll et al. (2012) argue that bilingualism creates optimal conditions for new language learning. In the same line, Kaushanskaya and Marian (2007, 2009) report an advantage for bilinguals over monolinguals with respect to new FL vocabulary learning. However, Bogulski and Kroll (in Kroll et al., 2012) demonstrated that when bilinguals have to learn a new language altogether, they do not show much difference in their performance to monolinguals learning the same L2, because bilingual experience includes the inhibition of the L1, not the L2 and accordingly, they cannot benefit from this previous experience with the inhibitory strategy (cf. also Green, 1998). In other words, when bilinguals learn the new vocabulary via the L1 or dominant language, they may apply the inhibitory strategy to their L1 and thus might outperform monolinguals, who do not have this strategy, but when learning proceeds via the L2 or non-dominant language of the bilingual, then this inhibition strategy does not apply and differences between bilingual and monolingual learners might be ruled out. This is likely to happen in classroom contexts where schooling proceeds in the majority language, for example where immigrant children coexist with local children. In this sense, the question is how this ability to make cross-linguistic comparisons affects their CLI behavior. For instance, this might help them avoid instances of negative CLI, and increase cognate use. On the contrary, recourse to L1/Ln has been shown to be a useful compensatory strategy, so one could assume bilinguals would trace back to their previous linguistic knowledge more often to compensate for lack of lexical knowledge. It is in our interest to disentangle this issue. With all these previous considerations in mind, the present study is intent on answering the following research questions: 1. What is the role of CLI in a lexical availability task? 2. What is the nature of CLI in a lexical availability task? 3. How do bilingual EFL learners differ from monolingual EFL learners in their CLI/lexical transfer behavior?

3 Study Design 3.1 Participants A group of 86 EFL learners participated in the study. They attended 2nd year of Baccalaureate (grade 12) a pre-university course, so at testing time they were between 17 and 18 years old. They all spoke Spanish as a mother tongue.

CLI in Lexical Accessibility

55

An additional group taken from a similar sample of students with bilingual learners also participated in the study. These EFL learners also speak another home language. This second student cohort is made up of 9 learners with either Romanian (5), Arabic (1), Portuguese (2), or Basque (1) as an additional language. All learners attended the same course, year 12, and a similar EFL program at the low intermediate level. The results of an Oxford Placement Test revealed they were in a low B1 level.

3.2 Instruments Data were collected using a lexical availability task. Learners were asked to complete a lexical availability task in English, which consisted of 15 prompts or centers of interest: parts of the body, clothes, the house, black and white, food and drink, make, animals, sad, school, the town, the countryside, love, professions, hobbies and hate. Learners had two minutes to write as many words as possible that came to their mind for each of the 15 stimuli or prompts, which makes up for 30 minutes allotted to complete the task. The main objective of the task was to find the words the learners had most available in their minds when prompted according to specific semantic fields without any restrictions as concerns formal or semantic relations. The task was completed in pen and paper form. The use of this task for this type of studies represents a novelty in methodology, since there are very few previous studies examining lexical access and lexical CLI through the lexical availability task.

3.3 Data Analysis and Procedures Once learners had completed the pen-and-paper lexical availability task, answers were transcribed into an Excel document. Responses were copied literally and uncorrected so as to allow for the identification of CLI episodes. Specifically, the following types of CLI instances with Spanish-L1-base were distinguished: • Phonetic spelling, learners write the L2 words , as they pronounce them, as in (1): (1) estomak for stomach,

• Lexical creation, learners create or coin a word adapting an existing L1 word to the morphosyntactic rules of the L2, as shown in (2): (2) noticies for news from noticias [Sp.].

56

M. del P. Agustín-Llach

• Literal translation, learners calque an expression form the L1 and literally render it into the L2 without any further consideration of possible restrictions, as illustrated in (3): (3) common time for time together from tiempo en común [Sp.]. (4) shared moments from tiempo en común [Sp.].

• Semantic extension, learners extend the semantic repertoire of an L1 equivalent to those (semantic) contexts where it can occur in Spanish, but not in English, as in (4): (5) police for policeman/policewoman from policía [Sp.] (both the profession and the abstract entity).

• Cognate use, learners use numerous instances of cognates, that is words which share form and meaning in the L1 and L2 such as in (5–7): (6) profession [Eng. Sp.: profesión], (7) doctor [Eng. and Sp.] or (8) tractor [Eng. and Sp.].

For descriptive and inferential statistical analysis, SPSS 19.0 was used. Nonparametric tests of means comparison were conducted, since the data sets did not meet the normality assumption. Specifically, Wilcoxon signed rank tests were conducted. To check for statically significant differences between the two independent samples, Mann-Whitney tests were also carried out.

4 Results This section will give account of the results obtained for the research questions posed above. Our first research question asked about the role of CLI instances in the total production of lexical items. Table 1 offers the results of the mean total production of lexical items in the lexical availability task, the mean total production of CLI types, and the relation between both. According to the results presented in Table 1. Around 15% of the total available words or lexical items produced in the lexical availability task include some kind of CLI of the types described in Sect. 3.3, including instances of cognate use. When CLI has a negative impact, the resulting lexical item is a lexical error. When cognates are used no error is present, and positive CLI is at stake.

Table 1 Mean number of total words and CLI instances produced per participant

Total CLI instances

Total words

Relationship CLI/words

32

214

32/214 × 100: 15%

CLI in Lexical Accessibility

57

Table 2 Mean results for CLI minus cognates and for total words available Total CLI instances minus cognates

Total words

Relationship CLI minus cognates/words

5.32

214

5.32/214 × 100: 2.5%

Table 3 Mean production of CLI occurrences Phonetic spelling

Lexical creation

Literal translation

Semantic extensión

Cognates

1.34

2.02

1.07

0.9

26.8

Accordingly, if cognates are subtracted from the CLI instance counts, results throw around 2.5% of CLI presence in the available lexical items produced. Table 2 offers this result. From this result, it follows that that the presence of (negative) CLI is relatively scarce. The type of task together with the level of the learners might account for this finding, as will be seen later. Exploration of the nature of CLI instances was the aim of the second research question. CLI occurrences were classified into the five types explained in Sect. 3.3. The mean number of occurrences for each CLI type are presented in Table 3. Mean values show that the most frequent type of CLI is cognate use, with an overwhelming superiority of production. This might be a mere reflection of the lexical systems of the target and source language. The second most frequent type of CLI is lexical creation, followed by phonetic spelling and literal translation. The extension of semantic features appears as the least frequent category. The type of task devoid of context might also help to account for these results. Lexical creations tend to be increasingly popular as learners’ proficiency goes up. Production of the different categories of CLI was checked for statistical differences in their frequencies of appearance. Accordingly, non-parametric tests of means comparison were conducted, specifically the Wilcoxon signed rank test, since the data did not meet the normality assumption. Results show statistical differences between the production of lexical creations and the rest of the categories and among cognates and the rest of the categories. The production of the remaining categories is not different in statistical terms. Table 4 displays the results of statistical tests. Finally, our third and last research question wanted to explore the differences between the monolingual and bilingual groups. CLI instances were analysed in both groups and comparisons established. Table 5 offers the data for the bilingual group and Table 6 presents the results of both groups. Table 5 presents the data for the bilingual participants with cognates being the most numerous type of CLI followed at distance by semantic extension and lexical creation. Phonetic spellings are the fourth most frequent type and literal translations the least frequent type of CLI.

58

M. del P. Agustín-Llach

Table 4 Statistical results for the two-way comparisons of frequencies of CLI types, Wilcoxon W, p values between brackets

Phonetic spelling

Phonetic

Lexical

Literal

spelling

creation

translation Extension

-

2.77 (.006) 1.17 (.243) 1.59 (.112) 8.05 (.000)

Lexical creation

-

Literal translation

Semantic Cognates

3.80 (.000) 4.71 (.000) 8.05 (.000) .345 (.730) 8.05 (.000)

-

Semantic extension

8.05 (.000)

-

Cognates

-

Table 5 Mean number of CLI types per bilingual participant Phonetic spelling

Lexical creation

Literal translation

Semantic extension

Cognates

0.77

1

0.33

1.11

25

Table 6 Comparison between the production of CLI in monolingual and bilingual learners

Monolingual EFL learners

Bilingual EFL learners

Mean tokens

214

216

Phonetic spelling

1.34

0.77

Lexical creations

2.02

1

Literal translations

1.07

0.33

Semantic extension

0.9

1.11

Cognate use

26.8

25

In Table 6 mean numbers for each CLI type for monolingual and bilingual groups are compared. Mean numbers show that the CLI behaviours are different for both groups, with the exception of recourse to cognates. In order to check statistical significance, non-parametric tests of mean comparison for two independent samples were conducted again, specifically Mann-Whitney’s u. The results obtained showed no significant differences for any of the categories. Table 7 shows the results of the statistical tests with the numbers for the MannWhitney’s u and significance values between brackets.

CLI in Lexical Accessibility Table 7 Results for the Mann-Whitney test, Mann-Whitney u

59

Phonetic spelling

u Sig. (2-tailed)

0.259

Lexical creations

u

242.5

Sig. (2-tailed)

0.060

u

268.5

Literal translations

302.5

Sig. (2-tailed)

0.102

Semantic extension

u

323.5

Sig. (2-tailed)

0.389

Cognate use

u

378

Sig. (2-tailed)

0.909

5 Discussion In the present section, results are interpreted and discussed in light of previous research-related studies. Our first research question wanted to explore the role of CLI in lexical items over the total lexical production. Results reveal two main situations. First, when cognates are included in the tallies of lexical interconnections, presence of lexical CLI amounts up to 15% over total lexical items. However, when cognates are withdrawn from the count, this number goes down to 2.5%. These results clearly reflect the great presence of cognates between English and Spanish with circa 3000 cognates (Meara, 1983). Actually, Nation and Meara (2002) contended that almost every Anglo-Saxon Word has a parallel Greco-Latin equivalent. In this sense, our results reflect real cognate presence in the FL and might concur with Otwinowska’s (2016, p. 86) statement that “cognate recognition and use is a natural strategy in learning FL”. When cognates are not considered, lexical CLI is 2.5% over total. This number is very low if compared with previous studies that found numbers of around 7% for learners in grade 6 where data were obtained from free writings (cf. Agustín-Llach, 2011). Two main reasons can account for this low number. First, the nature of the task might not promote CLI, since it is not a task of spontaneous communication. Moreover, although it is a time-constrained task, learners are not required to write a minimum number of words, so the linguistic and cognitive pressure is small. In this sense, results suggest that the lexical availability task partially inhibits L1/Ln influence. Second, the intermediate proficiency level of learners might also be playing a part in the low numbers of lexical CLI, since as learners get more proficient, reliance on the L1 to fill lexical gaps diminishes (e.g., Celaya & Ruiz de Zarobe, 2008; Ringbom, 2006). However, previous findings also reveal that learners at different levels of proficiency rely on different strategies to compensate for lack of lexical knowledge. In this sense, lexical creations become a more frequent strategy as learners gain mastery in the L2 (cf. Celaya & Ruiz de Zarobe, 2008; Dewaele, 1998; Ringbom, 2006). Literacy skills and school experience together with learners’ age might be determinant,

60

M. del P. Agustín-Llach

since younger learners at lower levels of schooling were found to transfer extensively in a lexical availability task (cf. Jiménez Catalán, personal communication). What these results are clearly putting forward is the evidence of some degree of simultaneous L1 + L2 activation (cf. Ecke, 2008; Jarvis, 2009; Levelt, Willem, Roelofs, & Meyer, 1999), thus both languages are activated even when only one language is required, which indicates parallel activity and cross-language competition (Kroll et al., 2012), and thus might suggest at least a timid mediating role of L1 during lexical access and retrieval. Additionally, our results show that even at intermediate levels of proficiency and language experience, lexical CLI persists (cf. Kroll et al., 2012). However, learners at this level are already quite competent and successful in suppressing cross-language associations (CLI). They manage to restrict word search primarily to the target language (at least under this condition where they did not experience much time pressure). Our second research question wanted to look into the impact of the different categories of lexical interconnections. The results clearly point to a notable presence of cognates, significantly more than any of the other categories. Our results support Ecke’s Parasitic Model of Vocabulary Acquisition, which focused on learners’ detection and use of similarity between new and already known words (see Ecke, 2015; Hall & Ecke, 2003). As already stated above, cognate use might reflect the real presence of cognates between the target and source language. But it might also be the result of a natural strategy or default learning mechanism of L2 learning (cf. Otwinowska’s, 2016), which Hall and Ecke (2003) called “parasitism”. In this sense, cognate use might be the result of (a) successful lexical creation or adaptation: e.g. intestine, abdominal, vacation, agriculture, (b) words learned and stored as L2 words without L1 or cross-linguistic identification: animal, director. In this sense, students simply know the word at stake (learned interlingual identifications, Jarvis, 2009): cathedral, stomach, architect or (c) the conscious or unconscious application of a compensatory or communicative strategy. The second main category of lexical CLI is lexical creations, which also throws significantly higher means than the rest of the categories. The most plausible explanation for the presence of lexical creations is the conscious application of a compensatory strategy due to the lack of lexical knowledge. Thus, the learner uses an L1 word base and tailors it to make it sound/look English, when he/she does not know the L2 word. Maybe, intermediate learners have enough knowledge about morphology and word formation regularities, which allow them to produce those lexical creations. Phonetic spellings come in the third most frequent position. They imply writing the lexical item as it is/would be pronounced in the L1. Accordingly, learners write fisics for physics, piyamas for pijamas, pensol for pencil. This reveals partial lexical knowledge, since the word is correct, but wrongly spelled. In addition, it might reflect interfering access routes or processing routines. Hall and Ecke (2003) refer to this as PCLI (performance CLI) as opposed to CCLI (competence CLI). Literal translations imply learners’ assumption of similar syntax, for example in our data, there are instances of altered order noun + adjective or noun complementation such as Bear polar, center-commercial, or program-television for television

CLI in Lexical Accessibility

61

program which has its base on L1 morpho-syntactic patterns. These calques alternate with correct word order such as alcoholic drink, wild animal, tourist guide. The least frequent category is semantic extension. Here, learners assume the semantic and contextual equivalence of the lexical items in the L1 and L2 such as in plate, career, note, professor (for teacher in the semantic field of School). Learners might have learned L2 lexemes and attached L1 lemmatic (semantic and syntactic) information to them (cf. Ecke, 2015; Hall & Ecke, 2003; Hall et al., 2009; Jiang, 2000 and their Parasitic Model). Accordingly, the lemma construction process is short-circuited in these cases. What the presence of these instances of lexical interconnections reveal is that the simultaneous activation of L1 and L2 is possible (albeit not very frequent) in the lexical availability task with intermediate-level L2 learners. Lack of lexical knowledge or faulty lexical selection are brandished as possible reasons for this dual activation (cf. Costa, Colomé, & Caramazza, 2000). Finally, our last research question asked about possible differences in lexical CLI behaviors of monolingual and bilingual learners learning English as an additional language. Results clearly show lack of significant differences in the mean number of instances of lexical interconnections produced by monolingual and bilingual learners. Our results, concur with Bogulski and Kroll (in Kroll et al., 2012) in that monolingual and bilingual learners learning a new L2 show similar levels of performance, since bilingual learners cannot apply the inhibitory strategy for the L2. In contrast, the results of the present study do not match the ones from Kaushanskaya and Marian (2007), who observed that bilinguals have a superior inhibitory control mechanism, what implies advantages when learning and using the L2. Bilinguals were found to retain more new foreign words and to have higher speed of retrieval. This might be timidly shown in our data in the mean number of tokens produced, which is slightly superior for the bilingual group. For Kaushanskaya and Marian (2007) bilinguals advantage in FL learning derives from their superior phonological skills including their phonological working memory. The lower numbers of phonetic spellings in the bilingual group might find an explanation here. However, differences are not significant and this remains as mere speculation based on the descriptive data, which might only be true for our specific data sets, and cannot be generalized in any case.

6 Implications for the FL Classroom From the above results concerning lexical interconnections, some pedagogical implications can be drawn to foster vocabulary teaching and learning in the EFL classroom. Evidence from lexical CLI calls for explicit lexical instruction of the spelling and pronunciation interface, including sound discrimination. This would help learners develop a more accurate spelling and abstract phono- graphemic patterns and associations. Lack of L2 writing experience can be at the origin of phonetic spellings, together with the prevalence of face-to-face or oral communication skills over written ones prioritized by the communicative approach in classroom teaching. Extensive

62

M. del P. Agustín-Llach

practice of writing in the L2 classroom is called for. By practicing writing learners are expected to improve their general language skills, and in particular, their writing of lexical items. Learners would benefit from a structured or directional approach to spelling and pronunciation, as explicit instruction would raise orthographic and phonetic awareness. Opportunities should also be provided to practice writing and saying new unfamiliar words in the English classroom. Cross-linguistic comparisons are also necessary to improve learners’ lexical competence. Therefore, the elaboration of learner-friendly teaching proposals based on similarities and differences of language pairs (L1/Ln-TL) should be in the teachers’ agenda. This comparison of L1–L2 lexical systems may prove very useful in L2 vocabulary instruction. Learners draw on their previous linguistic knowledge, usually L1, but also L3, or any other previous linguistic knowledge to compensate for deficiencies in their lexical knowledge (Odlin, 1989). As a consequence, learners’ vocabulary learning, recall and use can be improved from paying attention to the similarities and differences between the L1 and L2 lexical systems (Swan, 1997). In this sense, the use of cognates and other facilitative similarities can be beneficial to the learners. Teachers need to raise awareness of cognates and cross-linguistic similarities, as well. Thus, learners need to be trained to recognize cognates and capitalize on cross-linguistic similarities, enhance cross-linguistic awareness to explicitly and tacitly realize the links between languages and finally, use cognate vocabulary as a learning strategy. Learners do detect similarity and use it in parasitic fashion as an effective learning mechanism. What they do not seem to detect are the fine-grained differences. These need to be taught. Thus, learners need to be instructed about the semantic and pragmatic restrictions of L1 false cognates (e.g., equip is not equipo/team). These measures will contribute to overcoming inhibition and avoidance. The focus must be on the fostering and training into the (use of) communication and compensatory strategies and thus, help learners deal with vocabulary problems and lexical gaps. According to Martín Leralta (2007) and Salazar Campillo (2006), a strategy training program that is to be effective needs to include strategy application activities and strategy awareness raising. Previous research has highlighted the importance of systematic explicit vocabulary instruction and teaching of word learning strategies including strategies to enhance word learning, word memory and word recall (see Paviˇcic, 2008). Helping learners to develop their strategic competence by either teaching them new learning strategies or by reinforcing the use of existing ones is one of the best ways to contribute to successful L2 vocabulary acquisition (Paviˇcic, 2008). Thus, echoing previous studies (e.g., Council of Europe Threshold Level, 2001; Martín Leralta, 2007; Oxford, 2011), our proposal is a stage-wise vocabulary strategy training program consisting of three main phases, which can be a good model to train strategic competence in the FL classroom. These phases should account for development of students’ awareness of strategic competence, modelling of strategic behaviour on part of the teacher, student practice aimed at fostering autonomy, and a final step which includes student self-evaluation of the strategies used and potential transfer of the new strategies to new tasks (cf. also Chamot, 2004). Accordingly, in the first

CLI in Lexical Accessibility

63

phase, learners should pay attention to the demands of the task, set goals, plan how to address them, and activate existing knowledge. Teachers should collect information about the strategies that students use when learning and using vocabulary. Then they should raise students’ awareness about their own lexical learning processes in the FL. Asking learners questions aimed at getting this strategy-related information can be a good way to start the discussion and collect the information. In the second phase, teachers exemplify and illustrate the strategy. Students are asked to practice the strategy first in a teacher-controlled way in different activities which end in free practice. Finally, the last phase collects the strategic reflection and evaluation. It includes making judgments of value about outcomes and effectiveness of strategies (cf. Oxford, 2011, p. 25). Encouraging students to ask themselves questions about the efficacy of the strategies employed can be a good way to get them reflect about this and evaluate their strategic competence.

7 Conclusion The present study has intended to examine the relationship between lexical access, lexical knowledge, and transfer or cross-linguistic influence in the realm of word use. The goal of the present work was to examine instances of lexical CLI in a lexical availability task. Cognates made up for the biggest category of lexical CLI, but other less frequent types are also present. Among them, lexical creations are the most common, followed by phonetic spellings, literal translations and semantic extensions to comparable levels. The specific nature of the lexical availability task might explain the scarce presence of CLI instances, except for cognates. However, some instances of CLI could be attested, and thus conclude that the activation of the L1 and L2 during completion of the task is probable. Finally, no significant differences in the production of lexical CLI between monolingual and bilingual learners were found. The idea that bilinguals do not benefit from an L2 inhibitory strategy in the same way they do benefit from this strategy in the L1 might account for this. In the light of the lexical CLI instances, some pedagogical implications were recommended that might help improve learners’ lexical competence. Future lines of research should include the examination of other possible manifestations of CLI such as the use of phrases, use of collocations, words of different frequency levels, words of higher CEFR levels, and specific “non-shared” vocabulary. These CLI manifestations might be an insightful source to explore the relationship between L1 knowledge and L2 lexical access.

Cross-Linguistic Influence at the Level of Word Order in L3 English by L1 Georgian/L2 Russian Speakers Gvantsa Jichoshvili and M. Juncal Gutierrez-Mangado

Abstract This study investigates L3 transfer effects at the level of word order focusing on the OV/VO parameter (Neeleman & Weerman, 1999) in main and embedded clauses in three typologically unrelated languages: Georgian (OV), Russian (VO) and English (VO). Data were collected from three groups of English learners: L1 Georgian children, L1 Georgian adults and L1 Georgian/L2 Russian bilingual children (n = 53), all of them with a proficiency level of A2. The instruments used involved a grammaticality judgment task (GJT) and a translation task (TT) with affirmative and negative stimuli. The results from between group analyses showed high accuracy rates in all learners and no statistically significant differences on either task between the groups. This result indicates that at A2 level neither the monolingual nor bilingual learners show signs of transfer from their L1 Georgian or L2 Russian in the case of the latter. This finding is relevant for EFL teachers since it shows that already at an A2 proficiency level monolingual and bilingual learners have acquired the word order pattern of the target language, indicating that word order is not a source of difficulty in this particular context. However, within group analyses revealed that in the GJT task the L1 Georgian learners were not as good at rejecting ungrammatical sentences as accepting grammatical sentences. The L1 Georgian/L2 Russian learners, on the other hand, were equally accurate. This result may suggest that the bilingual learners are benefiting from their multilingual knowledge, which could have important implications for classroom practices. Keywords CLI · Word order · L1 Russian and Georgian · L3 English · Bilingual advantage G. Jichoshvili (B) University of the Basque Country UPV/EHU, Unibertsitateko Ibilbidea 5, 01006 Vitoria-Gasteiz, Spain e-mail: [email protected] M. J. Gutierrez-Mangado Department of English and German Philology and Translation and Interpreting, Faculty of Arts, University of the Basque Country UPV/EHU, Unibertsitateko Ibilbidea 5, 01006 Vitoria-Gasteiz, Spain e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_4

65

66

G. Jichoshvili and M. J. Gutierrez-Mangado

1 Introduction Studies on cross-linguistic influence (CLI) between three and more languages have emerged rather recently (Cenoz, 2001; Dewaele, 1998; Flynn, Foley, & Vinnitskaya, 2004; Hammarberg, 2001; Ringbom, 1987; Rothman, 2010 among others). One of the most widely discussed questions in the literature on third language (L3) acquisition is the effects that CLI can have on learner’s progress in mastering a new language. These effects can be facilitating or inhibiting (Odlin, 1989). Facilitation effects are generally known as positive transfer, whereas inhibition is considered as negative transfer. Positive transfer occurs when a particular grammatical feature or structure is found in the source language [e.g., first language (L1) or second language (L2)] as well as the target language (e.g., the L3) and can thus be the object of transfer and facilitate the acquisition process. Negative transfer refers to the transfer of different or similar but not identical features from the L1 and/or the L2 into the L3, leading to incorrect representations in the latter. In other words, while similarities in syntactic structures and vocabulary can facilitate the acquisition of grammar and reduce the time needed to develop the L3 (e.g., good reading comprehension), different L1 and L2 word order patterns may cause misinterpretation at different levels (Chapetón, 2008). Initially, CLI studies mainly focused on transfer effects in L2 acquisition, where the direction of transfer usually came from the L1 to the L2 (Gass, 1984; Kellerman, 1977; Odlin, 1989; Ringbom, 1987; Schwartz & Sprouse, 1996). However, the growing interest in multilingualism has resulted in studies investigating CLI involving more than two languages, where transfer can occur in different directions (Cenoz, 2001; Dewaele, 1998; Flynn et al., 2004; Hammarberg, 2001; Ringbom, 1987; Williams & Hammarberg, 1998): (1) from the L1 to the L3 (Bouvy, 2000; Leung, 2006); (2) from the L2 to the L3 (Bardel & Falk, 2007; Hammarberg, 2001; Rothman & Cabrelli Amaro, 2010); (3) from both the L1 and the L2 to the L3 (Flynn et al., 2004; Westergaard, Mitrofanova, Mykhaylyk, & Rodina, 2017); (4) from the L1 or the L2 to the L3 (Giancaspro, Halloran, & Iverson, 2015; Rothman, 2010); (5) from the L2 to the L1 (Cook, 2003; Kecskes & Papp, 2000), or (6) from the L3 to the L1 and the L2 (Lammiman, 2010).

2 Factors Affecting the Direction of CLI A number of factors, separately or in combination, have been proposed to determine the different directions of transfer: language proficiency in any of the languages involved (Ringbom, 2001; Williams & Hammarberg, 1998), the L2 status factor (Bardel & Falk, 2007), language distance and structural similarity (Rothman, 2011, 2015; Singleton, 1987; Westergaard et al., 2017), recency of use (Hammarberg, 2001) among others.

Cross-Linguistic Influence at the Level of Word Order …

67

With respect to language distance, some studies have reported CLI from typologically similar languages which share some type of structural similarity (De Angelis, 2005; Poulisse, 1999; Rothman, 2011, 2015; Westergaard et al., 2017), while others have also found CLI effects from typologically unrelated languages (Cenoz, 2001; Gutierrez-Mangado & Martínez-Adrián, 2018; Hermas, 2014; Martínez-Adrián, 2010; Martínez-Adrián, Gallardo-del-Puerto, & Gutierrez-Mangado, 2013; Rocca, 2007). This study investigates CLI at the level of word order in typologically different languages in three learner populations: one group of L1 Georgian child and adult learners of L2 English and one group of L1 Georgian/L2 Russian child learners of L3 English. The acquisition of word order is one of the well-researched topics in the L2 and L3 acquisition literature (Chan, 2004; Döpke, 1998; Meriläinen, 2010; Müller, 1998; Rothman, 2011; Rothman & Cabrelli Amaro, 2010; Westergaard et al., 2017; Yip & Matthews, 2000). According to Thomason and Kaufman (1988, p. 55), word order “seems to be the easiest sort of syntactic feature to borrow or acquire via language shift”. The results emerging from these studies have been diverse. Bardel and Falk (2007) compared the acquisition of sentential negation placement in two groups of learners. The first group was made up of five learners of L3 Swedish, whose L1s were Dutch (a V2 language), English or Hungarian, while their L2s were English, Dutch or German (also a V2 language). The second group was made up of four learners of L3 Dutch and L3 Swedish, both L3s being V2 languages. The only L3 Swedish learner had Italian as an L1 and German and Dutch as L2s. The remaining L3 Dutch learners’ L1s were Swedish or Albanian, while their L2s were English or German. All the learners were absolute beginners and the learning was formal. The results showed that only the group with a V2 L2 correctly transferred this structure into the L3. Those learners whose L1 was a V2 language did not show positive transfer of this feature into the L3. The results were interpreted as evidence for the L2 Status Factor, independently of typology. Angelovska and Hahn (2012) investigated the L3 English proficiency levels (A2, B1, B2, and C1) at which L2 negative transfer occurred in a group of 20–25-year-old learners with various L1s (Russian, Polish, Bulgarian, Croatian, Ukrainian, French, and Portuguese) and a constant variable of L2 German. The results of a written assignment (free text production) reported L2 negative syntactic transfer not only across learners from different L3 levels (with the exception of C1 level) but also in many syntactic forms and variations depending on the L3 proficiency level. Simsek (2006) investigated CLI at the level of word order in the written and spoken production of 14 Turkish-German bilingual learners of L3 English (aged 13–14) living in Germany and 25 L1 Turkish learners of L2 English (aged 13–15) living in Turkey. Turkish is an SOV language, whereas German is a V2 language with a relatively free word order (e.g., verb-first position in interrogative sentences, verb-second position in main clauses and verb-final position in subordinate clauses in declarative sentences). The results of the written production showed that the bilingual Turkish-German learners applied word order rules of German rather than Turkish, whereas the monolingual Turkish learners were influenced by Turkish word order features in their English sentence formation. These results were attributed to the

68

G. Jichoshvili and M. J. Gutierrez-Mangado

psychotypology effect, which resulted in German being perceived as sharing more grammatical features with English than Turkish. The authors also attribute the results to the effect of exposure and use in the sense that the bilingual subjects had more exposure to German than Turkish. Westergaard et al. (2017) investigated the acquisition of adverb-verb word order and subject-auxiliary inversion by 11–14-year-old Norwegian-Russian bilingual and Norwegian and Russian monolingual learners of English. Norwegian and English are typologically related in that they are both Germanic languages, while Russian is more distant from English. However, Norwegian is a V2 language, while English and Russian are not, although English shows residual V2 effects in subject-auxiliary inversion in questions. Thus, the researchers expected a facilitative effect from Russian in the judgements of adverb-verb word order in English, while at the same time they also expected a facilitative effect from Norwegian, since it also has subject-auxiliary inversion (while Russian generally does not). The results of the grammaticality judgement task showed facilitative CLI from both previously learned languages: the bilinguals’ Russian facilitated accuracy rates in the judgement of adverb-verb word order sentences when compared to the L1 Norwegian learners, who obtained lower scores. At the same time, the bilinguals’ Norwegian facilitated the correct judgement of subjectauxiliary inversion in questions when compared to the judgements of the L1 Russian learners of L2 English. However, the researchers also reported a non-facilitative effect from Norwegian in that the bilinguals rejected grammatical adverb-verb word order sentences more often than the L1 Russian participants. The authors ascribed this finding to what they refer to as the “double filter effect”, assuming that “all speakers have a grammaticality filter that is activated in cases where one has to process and judge linguistic data”. The number of filters is determined by the number of languages learners have acquired. Thus, before deciding on the grammaticality of a sentence in the L2, speakers analyse it by using the grammaticality filter of their L1, whereas bilingual speakers, at a stage when their L3 interlanguage is still unstable, consult the grammaticality filters of both previously acquired languages (Westergaard et al., 2017). To summarize, studies on CLI effects on word order in L3 acquisition have yielded mixed results reflecting positive or negative syntactic transfer from both the L1 and the L2 into the L3. Moreover, the majority of the studies compare the influence of two or three European languages (Foote, 2009; Giancaspro et al., 2015; Ionin, Montrul, & Santos, 2011), while the question of possible syntactic transfer effects between unrelated languages has but only been started to be investigated (Kulundary & Gabriele, 2012 with Tuvan/Russian/English; Özçelik, 2013 with Uzbek/Russian/Turkish; Wrembel, 2012 with Polish/French/English), and thus more research is needed in this respect. Thus, this study aims at identifying the possible transfer effects on word order between three typologically different languages: Georgian (Kartvelian), Russian (Slavic), and English (Germanic), a language combination which has not been investigated so far. Additionally, this study is one of a few studies that investigate CLI between typologically distant languages among child learners of L3 English (cf. Westergaard et al., 2017).

Cross-Linguistic Influence at the Level of Word Order …

69

Table 1 General differences between Georgian, Russian, and English Georgian

Russian

English

Type of language

Kartvelian

Slavic

Germanic

Orthography

Mkhedruli

Cyrillic

Roman

Highly inflectional

Yes

Yes

No

Basic word order

SOV

SVO

SVO

Preposition/postposition

Po

Pr

Pr

GN/NG

GN

NG

GN; NG

AN/NA

AN

AN

AN

Flexibility of word order

Flexible

Flexible

Rigid

Adapted from Greenberg (1966) SOV subject-object-verb; SVO subject-verb-object; Po postpositions; Pr prepositions; AN adjective noun order; NA noun adjective order; GN genitive noun; NG noun genitive

3 Word Order Patterns in Georgian, Russian, and English Georgian is a Kartvelian language, whereas Russian and English are Indo-European languages, although they fall into different branches (Balto-Slavic and Germanic branches, respectively) (Ethnologue Languages of the World: https://www. ethnologue.com/statistics/family). The different characteristics of these languages are summarized in Table 1. With respect to word order, as Table 1 shows, both Russian and English have an underlying SVO word order, while Georgian’s basic word order is SOV (Greenberg, 1966). As for Russian, Kallestimova (2007) showed that although it allows a freer word order than English, in thetic transitive sentences native speakers of Russian showed a clear preference for SVO word order (98.9%). Georgian, like Russian, also allows a variety of word orders but has been classified as a SOV word order language and has been reported to show a strong preference for SOV basic word order in thetic transitive sentences (Apridonidze, 1986; Aronson, 1982; Boeder, 2005; Harris, 2000; McGinnis, 1997; Pochkhua, 1962), the structures under study in the present chapter. Note that although the three languages are typologically different, with respect to word order, Russian shares more characteristics (word order) with English than Georgian does. In this respect, then, it could be claimed that Russian is more similar to English than Georgian is since they share the same underlying structural word order pattern (in the sense of Rothman, 2015). Below we provide the examples of sentences in Georgian, Russian and English with all possible word orders.

70

G. Jichoshvili and M. J. Gutierrez-Mangado

Georgian (1)

S

O

V

გოგონამ

ფაფა

შეჭამა.

gogona-m

pap-a

shecham-a

the girl-ERG.

the porridge-NOM.

ate-3sg.

S

V

O

გოგონამ

შეჭამა

ფაფა.

gogona-m

shecham-a

pap-a

the girl-ERG.

ate-3sg.

the porridge-NOM.

“The girl ATE the porridge”.

a

“The girl ate the porridge”. (2)

(3)

O

V

S

ფაფა

შეჭამა

გოგონამ.

pap-a

shecham-a

gogona-m

the porridge-NOM.

ate-3sg.

the girl-ERG.

“The porridge, it was the girl who ate it.” (4)

V

O

S

შეჭამა

ფაფა

გოგონამ.

shecham-a

pap-a

gogona-m

ate-3sg.

the porridge-NOM.

the girl-ERG.

“THE GIRL ate the porridge”. (5)

O

S

V

ფაფა

გოგონამ

შეჭამა.

pap-a

gogona-m

shecham-a

the porridge-NOM.

the girl-ERG.

ate-3sg.

“The porridge, the girl ATE it.” (6)

V

S

O

შეჭამა

გოგონამ

ფაფა.

shecham-a

gogona-m

pap-a

ate-3sg.

the girl-ERG.

the porridge-NOM.

“The girl did eat the porridge”. a The word order pa erns from (ii) to (vi) slightly change the meaning of the sentences. This is illustrated by the words in capital le ers in the glosses indica ng the informa on which is emphasized with the different word orders.

Cross-Linguistic Influence at the Level of Word Order …

71

Russian S (7)

V

O

Девочка

съела

кашу.

devochk-a

s’iel-a

kash-u

the girl-NOM.

ate-3sg.

the porridge-ACC.

S

O

V

Девочка

кашу

съела.

devochk-a

kash-u

s’iel-a

the girl-NOM.

the porridge-ACC.

ate-3sg.

O

V

S

Кашу

съела

девочка.

kash-u

s’iel-a

devochk-a

the porridge- ACC.

ate-3sg.

the girl-NOM.

“The girl ate the porridge.”

(8)

“The girl ATE the porridge.”

(9)

“The porridge, it was the girl who ate it.” V (10)

O

S

Съела

кашу

девочка.

s’iel-a

kash-u

devochk-a

ate-3sg.

the porridge-ACC.

the girl-NOM.

“THE GIRL ate the porridge.”

(11)

O

S

V

Кашу

девочка

съела.

kash-u

devochk-a

s’iel-a

the porridge- ACC.

the girl-NOM.

ate-3sg.

“The porridge, the girl ATE it.”

(12)

V

S

O

Съела

девочка

кашу.

s’iela

devochk-a

kash-u

ate-3sg.

the girl-NOM.

the porridge-ACC.

“The girl did eat the porridge.”

72

G. Jichoshvili and M. J. Gutierrez-Mangado

English (13)

(14)

S

V

O

The girl

ate

the porridge.

the girl-NOM.

ate-3sg.

the porridge-ACC.

*S

O

V

*The girl

the porridge

ate.

the girl-NOM.

the porridge-ACC.

ate-3sg.

Table 2 shows the different possible word order combinations allowed in Georgian, Russian and English. Table 2 Possible word order combinations in Georgian, Russian and English

Georgian

Russian

English

Basic word order

SOV

SVO

SVO

Other word orders

SVO OVS VOS OSV VSO

SOV OVS VOS OSV VSO

4 Research Questions and Hypotheses Given the word order preferences reported in the literature in thetic transitive sentences in Russian (SVO) and Georgian (SOV), the research questions entertained in this study are the following: RQ1: Do L1 Georgian/L2 Russian child learners of L3 English show any transfer effects at the level of word order at an elementary level? RQ2: Does CLI come from the language that shares the same word order (L2 Russian), leading to positive transfer or from the language with a different word order (L1 Georgian), leading then to negative transfer? More specifically, we entertain the following scenarios for the L1 Georgian/L2 Russian learners of L3 English: – The L1 Georgian/L2 Russian bilingual participants will be more accurate than their monolingual peers as well as the adults. This would be interpreted as a case of facilitative transfer from their L2 since Russian, like English, shows SVO word order with transitive verbs in main and embedded clauses. – The L1 Georgian/L2 Russian bilingual participants will be as accurate as their monolingual peers and adults. This result could be interpreted as non-facilitative transfer in the bilingual as well as the monolingual participants in case the accuracy

Cross-Linguistic Influence at the Level of Word Order …

73

rates are low in all groups. However, if accuracy rates are equally high for all groups, then, this would indicate that the learners may have already acquired the target language SVO word order.

5 The Study 5.1 Participants The data were gathered at a secondary school with intensive English language instruction and the Agricultural University of Georgia both in Tbilisi, Georgia. The participants were 53 students (32 males and 21 females) divided into the following groups: (1) 16 L1 Georgian child learners of L2 English (mean age = 11.75); (2) 19 L1 Georgian/L2 Russian bilingual child learners of L3 English (mean age = 11.68); and (3) 18 L1 Georgian adult learners of L2 English (mean age = 19.5) (Table 3). The child participants were at the sixth grade of secondary school and had started learning English in the second grade (mean age of first exposure 6.38), having received around 3 h and a half of EFL classes per week. The adult participants were undergraduate students and had started learning English at a mean age of 7.78. Most of them were first year students from the Agricultural University of Georgia and at the time of data collection (first semester) they were receiving 6 h of English per week. Both children and adults demonstrated to have the same A2 level of English.1

5.2 Materials The participants’ level of English was established through the Quick Oxford Placement Test, which placed them at an elementary A2 language level (Council of Europe, 2001).2 Level A2 was aimed at since previous studies have demonstrated that transfer at the word level mostly tends to occur at lower levels of proficiency (e.g., Rast, 2010). The participants also completed a linguistic background questionnaire, which allowed us to classify the participants into bilingual and monolingual learners. The materials used to elicit word order data involved (1) a grammaticality judgment task (GJT) and (2) a translation task (TT). GTJs have been widely used in 1 The

group of adults was included in the original study because we were interested in analyzing whether the adults and the children performed differently (i.e., produced more instances of transfer from L1 Georgian to L2 English), in line with other studies such as Flynn et al. (2004). However, due to space limitation, we will not discuss adult-child differences in the present paper. 2 It should be noted that irrespective of the fact that before entering university the adult learners were exposed to English in formal settings (age of exposure around 7 years), their level of English was no higher than A2. This can be explained by the students’ reluctance to study foreign languages at school or by them having or attending fewer English classes per week at school in comparison with the child participants.

74

G. Jichoshvili and M. J. Gutierrez-Mangado

Table 3 Participants Group

Number of participants (M = males; F = females)

Average age

First age of exposure to English

G1: L1 Georgian children

16 (6 M, 10 F)

11.75

6.38

G2: L1 Georgian adults

18 (16 M, 2 F)

19.5

7.78

G3: L1 Georgian/L2 Russian children

19 (10 M, 9 F)

11.68

6.37

research on CLI successfully in the past (Falk & Bardel, 2011; Martínez-Adrián, 2010; Rothman, 2010). With respect to the TT, previous research has shown that TTs tend to elicit CLI effects more easily than tasks that call for free composition (Ringbom, 1987) since they force the learner to produce specifically the target form (Moattarian, 2013).

5.3 Experiment 1: The Grammaticality Judgment Task (The GJT) The GJT was made up of 80 sentences, which were divided into eight groups in accordance with their word order and their level of embedding [main (M) and embedded (E)], and they all contained a subject (S), a verb (V), a direct object (O), and a prepositional phrase (PP) (15–22). The different word order combinations yielded 20 grammatical sentences, 40 ungrammatical sentences, and 20 fillers involving negation (see Appendix 1).3 Matrix sentences: VOPP: Subject + Verb + Object + Prepositional Phrase (15) The woman locked the door in the room.

OVPP: Subject + Object + Verb+Prepositional Phrase (16) *The boy the book read in the room.

OPPV: Subject + Object + Prepositional Phrase + Verb (17) *The monkey the banana in the garden ate.

Embedded sentences: VOPP: Subject + Verb + Object + Prepositional Phrase (18) The teacher thinks that the boy forgot the pen on the table. 3 Initially,

the GJT was composed of 120 sentences: 40 grammatical, 40 ungrammatical, and 40 fillers. However, later, as a precaution against fatigue, it was decided to cut the number of grammatical sentences and fillers from 40 to 20.

Cross-Linguistic Influence at the Level of Word Order …

75

OVPP: Subject + Object + Verb + Prepositional Phrase (19) *The bear thinks that the rabbit the carrot saved in the yard.

OPPV: Subject + Object + Prepositional Phrase + Verb (20) *The footballer thinks that the boy the ball in the yard caught.

Fillers: (22) *The boy thinks that the puppy not drink the milk. (21) *The footballer not win the game in the summer.

All 80 sentences were randomized and divided into two sets, with each containing 40 matrix (10 VOPP, 10 OVPP, 10 OPPV, 10 fillers) and 40 embedded (10 VOPP, 10 OVPP, 10 OPPV, 10 fillers) sentences. Each set was given to the each participant on a different day in order to avoid the loss of concentration and fatigue of the participants. All the sentences contained only transitive verbs in the past simple tense form where the subject and the object were always definite noun phrases (NPs). The participants were given a binary choice, where they had to indicate whether each sentence sounded “correct” or “incorrect” by circling the corresponding “smiley” (☺ = “correct”,  = “incorrect”). The learners were not given any time limit.

5.4 Experiment 2: The Translation Task (The TT) The TT consisted of 24 sentences in Georgian which the participants were asked to translate into English: 6 OV and 6 OPPV matrix sentences, and 6 OV and 6 OPPV embedded sentences (23–26). Each sentence type was represented by 3 sentences with animate and 3 sentences with inanimate objects all in Mkhedruli alphabet (see Appendix 2). The TT was administered in Georgian since this was the L1 for all participants. Matrix sentences: (23) OV:

ქალმა

გოგონას

დაურეკა

qal-ma

gogona-s

daurek-a

woman-ERG.

girl-DAT.

called

“A woman called a girl.” (24) OPPV:

გოგონამ

ოთახი

დილით

დაალაგა

gogona-m

otakh-i

dili-t

daalag-a

girl-ERG.

room-NOM.

morning-IN

cleaned

“A girl cleaned a room in the morning.”

76

G. Jichoshvili and M. J. Gutierrez-Mangado

Embeded sentences: (25) OV:

დედოფალი

ფიქრობს,

რომ

კატამ

თაგვი

დაიჭირა

dedopal-i

piqrob-s,

rom

kata-m

tagv-i

daichir-a

queen–NOM.

thinks

that

cat-ERG.

mouse-NOM. caught

“A queen thinks that a cat caught a mouse.” (26) OPPV: სპილო sp’ilo

ფიქრობს, რომ ვეფხვმა piqrob-s,

elephant-NOM. thinks

წყალი

rom vepkhv-ma tskal-i that

ger-ERG.

დილით

დალია

dil-it

dali-a

water-NOM. morning-IN drank.

“An elephant thinks that a ger drank water in the morning.”

5.5 Procedure Students were told to focus on the meaning of the sentences and to try to convey their meaning as accurately as possible. The vocabulary used in the GJT and the TT corresponded to the level of the vocabulary used in the participants’ school textbooks. The experiment was conducted in three days. On the first day, the participants were given the linguistic background questionnaire. On the second day, they were given the Quick Oxford Placement Test and the first part of the GJT, and on the third day, they completed the second part of the GJT and the TT.

5.6 Codification and Analysis The results were codified as follows: all responses in the GJT and the TT were coded and given a value (“1” for correct and “0” for incorrect). In the TT, the sentences with correct word order were counted as correct and given a value irrespective of the mistakes on verb forms or the incorrect use of definite and indefinite articles. As for statistical analyses, the data were analyzed using SPSS software. In order to check for normal distribution, we ran the Kolmogorov-Smirnov test, which showed that the data were not normally distributed for most conditions, except for the OV matrix (Z = .164, p = .200) and embedded (Z = .184, p = 0.152) and OPPV (Z = .133, p = .200) conditions. Therefore, non-parametric tests were used throughout the study.

6 Results In this section the results of the GJT will be presented first then followed by the results of the TT.

Cross-Linguistic Influence at the Level of Word Order …

77

6.1 Experiment 1: The GTJ Results With respect to matrix sentences, Table 4 shows the number of correct answers and their percentages of all three groups. The Kruskal-Wallis test for between-group comparisons showed no statistically significant differences for any of the types of sentences (VOPP z = 1.026, p = .599; OVPP z = 3.439, p = .179; OPPV z = 3.318, p = .190). With respect to the responses for the embedded sentences (Table 5), the KruskalWallis test for between-group comparisons revealed no statistically significant differences among the groups for any of different word order types (VOPP z = 1.050, p = .592; OVPP z = 2.446, p = .294; OPPV z = 4.892, p = .087). For within group comparisons, for Group 1, the Friedman test showed significant differences in accuracy rates for different sentences (z = 67.889, p = .000). The Wilcoxon test showed that VOPP sentences were judged correctly more often than any other sentences in matrix (OVPP z = −2.407, p = .016; OPPV z = −2.620, p = .009) as well as embedded sentences (OVPP z = −3.000, p = .003; OPPV z = −2.692, p = .007). No differences were found between any of the three types of sentences when matrix and embedded clauses were compared (VOPP z = .237, p = .813; OVPP z = .174, p = .862; OPPV z = 1.267, p = .205). For Group 2, similarly to Group 1, the Friedman tests revealed statistically significant differences in the number of sentences judged correctly (z = 46.797, p =

Table 4 Number, percentage and standard deviation (sd) of correct answers in matrix sentences Matrix

VOPP

OVPP

OPPV

Group

Maximum Correct number answers

% (sd)

Correct answers

% (sd)

Correct answers

% (sd)

G1

160

144

90 (1.26)

117

73.10 (2.35)

124

77.50 (1.61)

G2

180

164

91 (1.41)

145

80.50 (1.83)

156

86.60 (1.81)

G3

190

164

86.31 (1.31)

163

85.70 (1.83)

155

81.50 (2.67)

Table 5 Number, percentage and standard deviation (sd) of correct answers in embedded sentences Embedded

VOPP

OVPP

OPPV

Group

Maximum Correct number answers

% (sd)

Correct answers

% (sd)

Correct answers

% (sd)

G1

160

143

89.30 (1.06)

116

72.50 (1.69)

114

71.25 (1.85)

G2

180

162

90 (1.32)

137

76.10 (2.42)

150

83.30 (1.87)

G3

190

154

81.05 (2.49)

157

82.60 (1.69)

161

84.70 (1.71)

78

G. Jichoshvili and M. J. Gutierrez-Mangado

.000). The Wilcoxon test showed that they correctly accepted VOPP sentences more often than they correctly rejected OVPP matrix (z = 1.974, p = .048) and embedded (z = 2.021, p = .043) sentences. Also, similarly to Group 1, no differences were found between any of the three types of sentences when matrix and embedded clauses were compared (VOPP z = .359, p = .719; OVPP z = .828, p = .408; OPPV z = .960, p = .337). On the other hand, the Wilcoxon test showed no differences within Group 3 for any type of sentences (z = 1.999, p = .851).

6.2 Experiment 2: The TT Results The results of the TT are presented in Tables 6 and 7. The results of the KruskalWallis test showed no differences among the three groups for OV (matrix z = 2.144, p = .342 and embedded z = 1.896, p = .387), and OPPV (matrix z = .796, p = .672 and embedded z = .299, p = .861). No within-group differences were found in either group.

Table 6 Number, percentage and standard deviation (sd) of correctly translated OV matrix and embedded sentences OV matrix Group

Maximum number

Correct answers

OV embedded % (sd)

Correct answers

% (sd)

G1

96

95

98.9 (.25)

95

98.9 (.25)

G2

108

105

97.22 (.51)

108

100 (0)

G3

114

114

100 (0)

111

97.36 (.31)

Table 7 Number, percentage and standard deviation (sd) of correctly translated OPPV matrix and embedded sentences OPPV matrix Group

Maximum number

Correct answers

OPPV embedded % (sd)

Correct answers

% (sd)

G1

96

92

95.80 (.77)

94

97.90 (.34)

G2

108

105

97.20 (.7)

104

96.29 (.73)

G3

114

104

91.22 (1.07)

110

96.49 (.41)

Cross-Linguistic Influence at the Level of Word Order …

79

7 Discussion The analyses from the GJT and the TT revealed no statistically significant differences between the three groups, with all three groups correctly accepting the grammatical sentences and rejecting the ungrammatical sentences with equal high accuracy. In light of these results and in answer to our first research question (Do L1 Georgian/L2 Russian child learners of L3 English show any transfer effects at the level of word order at an elementary level?), the results showed that the L1 Georgian/L2 Russian bilingual children did not perform differently from the L1 Georgian monolingual children and adults on the GJT and the TT. This result indicates that the bilingual children do not transfer from their L2 Russian, since their responses are not different from their L1 Georgian peers and the group of adults. As for research question 2 (Does CLI come from the language that shares the same word order (L2 Russian), leading to positive transfer or from the language with a different word order (L1 Georgian), leading then to negative transfer?), the absence of differences in the judgements and translations among the three groups, together with the high accuracy rates in the GJT task (above 70%) and the TT task (over 95%) seem to indicate that there was no negative transfer from Georgian in any of the groups. All the participants correctly accepted the VOPP grammatical sentences with the same high accuracy rates, and they also correctly rejected the ungrammatical OVPP and OPPV sentences with the same accuracy rates. In the case of the bilingual group, given the absence of between-group differences, we cannot establish that there was any facilitative transfer from the typologically closer Russian. Therefore, our prediction that Russian would have a facilitative effect was not borne out, which could indicate that the learners in our study (the bilingual children as well as the monolingual children and adults) have already acquired the word order in English at the level of A2. As pointed out by one reviewer, the results we have obtained could be a confound of proficiency and exposure to the target language in the sense that the participants tested, although still at a low general level of proficiency in the target-language, may crucially be already ‘in transition towards or at the L3 target’ (Gónzalez Alonso & Rothman, 2017). Further investigations would be needed to test whether at true initial stages (e.g., first hours of exposure to the target language) learners do show some non-facilitative transfer effects from L1 Georgian and whether facilitative transfer from L2 Russian can be detected. However, the within-group comparisons show that both monolingual groups, Groups 1 and Group 2, were more accurate at accepting the grammatical VOPP sentences than rejecting the ungrammatical OVPP sentences, both in matrix and embedded contexts. In contrast, the bilingual group, Group 3, did not differ in the rates at which they accepted grammatical sentences and rejected ungrammatical sentences. In other words, the monolingual learners were better at correctly accepting VO sentences than rejecting OV sentences, while the bilingual group was equally good at accepting the grammatical sentences, as at rejecting the ungrammatical ones. In this respect, the bilingual group is more homogeneous in their judgments than the

80

G. Jichoshvili and M. J. Gutierrez-Mangado

monolingual group. This difference could indicate that the knowledge of two languages places the bilinguals at an advantage in the sense that they are equally good at correctly accepting grammatical sentences as correctly rejecting ungrammatical sentences. The monolingual learners, on the other hand, are better at accepting the grammatical sentences than rejecting the ungrammatical sentences, which might indicate that having only one language in their linguistic repertoire from which to potentially transfer (only one grammaticality filter active in Westergaard et al.’s terms), makes it more difficult for them to reject the ungrammatical sentences. In a way, although we did not find any between-group differences, the finding that the monolinguals were not as good at rejecting ungrammatical sentences might indicate some sort of weak CLI from L1 Georgian since they have more difficulties rejecting precisely sentences which display L1 Georgian word order. However, additional data would be needed to provide a more robust confirmation of this finding. The finding that the bilinguals are more homogeneous in their judgements than the monolinguals is in line with research that has shown that bi/multilingual learners acquire and process additional languages differently from monolinguals. Often, this difference tends to favour bi/multilingual learners (Cenoz, 2003c; McLaughlin & Nayak, 1989; Nation & McLaughlin, 1986). As reviewed in Cenoz (2003c), one of the reasons why bi/multilinguals have an advantage over monolinguals may be related to the fact that bi/multilinguals have already undergone the experience of learning an additional language. For monolinguals, on the other hand, learning an L2 is a ‘first’ experience. It has been suggested that the bi/multilingual language learners are able to apply the learning/processing strategies they developed when learning their L2 to the process of learning an additional language. Bi/multilingual learners have also been reported to show higher metalinguistic awareness than monolinguals (Bialystok, 1991, 2001; Bialystok & Barac, 2012; Cummins, 1978; Galambos & Hakuta, 1988; Lasagabaster, 1997; Ricciardelli, 1992 among others). In this respect, Bialystok and Barac (2012) suggest that metalinguistic tasks, such as a GJT, focus on linguistic representations, and that bi/multilingual learners, by virtue of knowing two languages, have a better knowledge of abstract linguistic representations, which results in enhanced metalinguistic awareness. We suggest that the findings reported in our study, showing that the L1 Georgian/L2 Russian child learners of L3 English, in contrast to the L1 Georgian child and adult learners of L2 English, were equally good at accepting the grammatically correct sentences as they were at rejecting the grammatically incorrect sentences may stem from their being experienced language learners and being metalinguistically more aware of the possible differences between languages than the monolingual groups. Note that both tasks, the GJT and the TT tap explicit knowledge, which can potentially be influenced by metalinguistic awareness. It may be the case that being metalinguistically aware of two different word orders (Georgian SOV and Russian SVO) allows them to be as accurate at rejecting the ungrammatical sentences as at accepting the grammatical ones.

Cross-Linguistic Influence at the Level of Word Order …

81

To summarize, we set out to explore whether the L1 Georgian/L2 Russian child learners of L3 English would show any facilitative transfer effects at the level of word order from their L2 Russian when compared to a group of L1 Georgian learners of L2 English matching in age and a group of L1 Georgian adult learners of L2 English, all at an A2 level of English. The results showed that the learners in all three groups performed very accurately, accepting the grammatically correct sentences and rejecting the ungrammatical sentences in the GJT and also accurately translating different word order sentences from L1 Georgian into English. This result cannot be interpreted straightforwardly as indicating absence of facilitative transfer effects from L2 Russian in the bilingual group since it may be the case that both the monolingual and bilingual learners have already moved on beyond the initial stage. As a reviewer pointed out, exposure and proficiency do not necessarily go hand in hand (Gónzalez Alonso & Rothman, 2017) so it may well be the case that, despite still being at a beginner level of proficiency, the participants in this study have already acquired the target language word order. However, the finding that the bilinguals, unlike the monolinguals, are equally good at accepting grammatical sentences as they are at rejecting ungrammatical sentences, suggests that the L1 Georgian/L2 Russian children may, after all, have benefited from their knowledge of an additional language in that their grammatical judgments were more homogeneous than the monolinguals’.

8 Pedagogical Implications The need to establish bridges between research findings in formal language acquisition studies and the teaching/learning of foreign languages has long been advocated for (Gil, Mardsen, & Whong, 2017; Gónzalez Alonso & Rothman, 2017; Leal & Slabakova, 2017; Lightbown, 2000; Martínez-Adrián et al., 2013; Nassaji, 2012). The flow of information between researchers and practitioners should be bidirectional, so that researches investigating language acquisition can benefit from the experience and actual difficulties teachers encounter in their increasingly multilingual classrooms. In their turn, foreign language teachers can benefit from knowledge about the differences that may arise at different levels in the learning of the target language. On the one hand, research on CLI has shown that certain target language features of English, such as phonetics/phonology (Cebrian 2009; Cenoz & GarcíaLecumberri, 1997a; Martínez-Adrián et al. 2013) and vocabulary (De Angelis & Selinker, 2001; Jarvis, 2009; Ringbom, 2001) are likely to induce negative CLI and, thus, may be harder to acquire. Similarly, research has also shown that the linguistic history of a group of learners of the same target language which also share their native language may have significant effects on the learning/acquisition process. More specifically, it may be the case that a subset of these learners speaks an additional language, which has been shown to be a potential source of differences in the classroom. In other words, being already a bilingual learner when first exposed to

82

G. Jichoshvili and M. J. Gutierrez-Mangado

a foreign language may trigger differences in the process of acquiring the foreign language when compared to learners whose sole prior linguistic repertoire is their native language (Gónzalez Alonso & Rothman, 2017; Jessner, 2006; Rothman & Cabrelli Amaro, 2010). Research can also inform about how and when monolingual and bilinguals learners will differ in their patterns of acquisition of the same target language. This way, knowledge about the potential areas of difficulty and differences in the language repertoires of their particular students would allow teachers to predict specific areas of difficulties and plan ahead and create pedagogical interventions leading to diminishing these potential differences. In this respect, the present study has shown that at the level of word order in matrix and embedded sentences, L1 Georgian and L1 Georgian/L2 Russian children do not seem to differ in their accuracy rates, which seems to indicate that the VO word order of English is not a source of difficulty for these groups of learners at least at an A2 proficiency level. Thus, EFL teachers whose students have this particular language combination can focus their teaching practice on other language features (which still need to be identified), which may prove more difficult for these particular learners. However, the results have also revealed a difference in the way monolingual and bilingual learners reject ungrammatical sentences. This finding may be of relevance for EFL teachers since it shows that despite no apparent differences between monolingual and bilingual children, bilinguals are more accurate at rejecting ungrammatical sentences in a GJT, which may be an indication of their higher metalinguistic awareness. Benefits of metalinguistic awareness on tasks that require analysis of linguistic knowledge such as detection of syntactic ambiguity and grammaticality judgements abound in the literature (Byalistok, 1991, 2001; Kemp, 2001; Mohanty, 1994; Renou, 2001; Thomas, 1992 among many others). However, it has also been suggested that it is necessary to explicitly teach learners about the differences/similarities between the languages in their linguistic repertoire in order for them to develop metalinguistical awareness (James, 1996; Thomas, 1988) and thus reap its benefits in language learning. The knowledge that monolingual learners have difficulties rejecting ungrammatical OV sentences in English may make teachers consider the possibility of designing class activities to explicitly highlight the fact that English and Georgian do not have the same basic word order and that while Georgian allows a variety of different word orders, English is more restricted and only allows VO word orders in main and embedded thetic transitive sentences. In this way, teachers would be raising the monolinguals’ metalinguistic awareness, which could potentially lead to higher rates of rejection of ungrammatical sentences in the monolingual learners. Acknowledgements The authors would like to acknowledge the grants awarded by the Spanish Ministry of Economy and Competitiveness (FFI2012-32212) and (FFI2016-74950-P) (AEI/FEDER/UE), the University of the Basque Country UPV/EHU (UFI 11/06) and the Basque Government (IT904-16).

Cross-Linguistic Influence at the Level of Word Order …

Appendices Appendix 1: Grammaticality Judgment Task (The GJT) Test A. Matrix Sentences (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37)

The boy fed the rabbit in the garden. The lion drew the mouse in the forest. The boy called the girl in the evening. The tiger met the elephant in the jungle. The girl found the frog in the box. The man bought the umbrella in the shop. The woman locked the door in the room. The girl needed the cream for the pie. The princess baked the cake for the party. The prince forgot the sandwich in the morning. The girl the mobile in the bag kept. The monkey the banana in the garden ate. The king the story to the princess told. The woman the mirror from the pocket took out. The queen the dress for the girl made. The girl the boy in the garden watched. The prince the butterfly in the garden saw. The princess the frog near the lake kissed. The child the bird in the cage left. The penguin the fish in the ocean caught. The boy the book read in the room. The girl the flower smelled for the sister. The mother the hamburger made in the evening. The father the piano played with the children. The boy the ball kicked in the yard. The girl the window broke in the classroom. The dog the door locked in the car. The zebra the water drank in the zoo. The child the toy lost in the garden. The princess the key put in the pocket. The boy not see the rabbit in the garden. The cowboy not ride the horse in the forest. The painter not draw the princess in the castle. The cat not catch the mouse in the kitchen. The boy not leave the puppy in the yard. The footballer not win the game in the summer. The builder not forget the box in the car.

83

84

G. Jichoshvili and M. J. Gutierrez-Mangado

(38) The student not know the answer to the question. (39) The girl not read the book in the classroom. (40) The boy not play the football in the yard. Embedded Sentences (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37) (38) (39) (40)

The princess thinks that the prince lost the sword in the yard. The mother thinks that the baby kicked the ball in the garden. The man thinks that the puppy closed the window in the room. The boy thinks that the girl made the cake for the party. The mother thinks that the girl cleaned the room in the morning. The puppy thinks that the kitten found the balloon in the garden. The dog thinks that the cat rode the bicycle in the kitchen. The tiger thinks that the elephant drank the water from the lake. The dog thinks that the cat climbed the tree in the jungle. The teacher thinks that the boy forgot the pen on the table. The bear thinks that the rabbit the carrot saved in the yard. The boy thinks that the footballer the game won in the evening. The bird thinks that the fox the egg stole from the nest. The queen thinks that the princess the dress bought for the party. The woman thinks that the girl the toy put in the box. The teacher thinks that the child the book read in the kitchen. The princess thinks that the cat the rat saw in the garden. The cat thinks that the dog the car drove in the evening. The horse thinks that the bear the apple ate in the forest. The doctor thinks that the girl the arm broke in the morning. The panda thinks that the giraffe the violin in the forest played. The cow thinks that the horse the orange in the morning wanted. The goat thinks that the duck the milk on the chair spilled. The mother thinks that the girl the dress in the morning washed. The king thinks that the princess the teddy bear in the yard left. The aunt thinks that the uncle the letter in the evening wrote. The footballer thinks that the boy the ball in the yard caught. The prince thinks that the princess the song in the castle sang. The zebra thinks that the monkey the banana in the jungle picked. The man thinks that the builder the house in the village built. The princess thinks that the cat not find the mouse. The doctor thinks that the boy not take the medicine. The boy thinks that the puppy not drink the milk. The girl thinks that the elephant not eat the grass. The queen thinks that the king not like the cake. The teacher thinks that the student not write the story. The boy thinks that the runner not lose the race. The mother thinks that the baby not watch the cartoon. The prince thinks that the princess not make the sandwich. The girl thinks that the bird not sing the song.

Cross-Linguistic Influence at the Level of Word Order …

85

Appendix 2: Translation Task (The TT) • SOV Matrix “A girl tickets bought”. “A footballer a game won”. “A monkey a banana ate”. “A boy a tiger saw”. “A girl a frog found”. “A woman a girl called”.

(1) (2) (3) (4) (5) (6) • SOPPV Matrix

“A girl a room in the morning cleaned”. “A father a keys in the yard lost”. “A girl a mobile in the bag put”. “A prince a horse in the yard left”. “A princess a puppy in the room left”. “A mother a butterfly in the garden saw”.

(1) (2) (3) (4) (5) (6) • SOV Embedded

“A boy thinks that a father a car

(1) bought”.

“A girl thinks that a

(2) mother a cake baked”.

“A granny thinks that a

(3) grandpa a letter wrote”.

“A prince thinks that a

(4) princess a king met”.

“A mother thinks that a child a

(5) kitten found”.

“A queen thinks that a cat

(6) a mouse caught”. • SOPPV Embedded

“A girl thinks that a

(1) boy a toy in the garden left”.

“An elephant thinks

(2) that a tiger a water in the morning drank”.

“A granny thinks

(3) that a girl a sandwich in the evening ate”.

“A boy thinks that a girl

(4) a bird in the yard left”.

86

G. Jichoshvili and M. J. Gutierrez-Mangado

(5)

“A woman thinks that a man a wolf in the forest saw”. “A

(6) thinks that a princess a rabbit in the garden fed”.

prince

Didactic Challenges in the Multilingual Classroom—The Case of French as a Foreign Language Christina Lindqvist

Abstract The present study explores the multilingualism of Swedish learners of French. More specifically, it looks at the use of multiple languages, manifested as lexical and grammatical cross-linguistic influence, in written retellings by Swedish pupils aged 12–16 (N = 105). In Sweden, formal learning of French in school starts in grade 6 or 7, that is around 3 years after the start of the formal learning of English. Thus, in this respect, French can be considered as a third language (L3) for most Swedish pupils. In addition to that, many pupils also study other foreign languages, usually Spanish or German, which means that they undergo learning processes in several languages simultaneously. Moreover, nowadays it is common for pupils to have knowledge of yet other languages spoken at home. Thus, it is clear that the learning of French by Swedish learners usually takes place under multilingual circumstances and therefore it can constantly be affected by other languages. While the participants in the present study all have Swedish as their L1, they know many other languages, to various degrees. The results show that some of these languages, especially those learned formally in school, are used when pupils write in French. The didactic implications of these results are discussed. What are the challenges for language teachers in multilingual classrooms, at present and in the future? Keywords Multilingual classroom · L3 French · Cross-linguistic influence · Written production

1 Introduction It is often mentioned within the field of research on second language acquisition, and, perhaps even more often, within research on third language acquisition, that multilingualism is becoming more and more widespread in the world (Hammarberg, C. Lindqvist (B) Department of Languages and Literatures, University of Gothenburg, Box 200, 40530 Gothenburg, Sweden e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_5

87

88

C. Lindqvist

2009). The reasons for this development are many, one of them being migration. It is evident that more and more languages are used in a society where people move between different countries. Another reason is the fact that pupils have the possibility to study more than one language formally in school in many countries. This is the case in Sweden, where pupils normally learn English as their first foreign language starting at the age of nine or ten, that is, in grade 3 or 4. Then, usually at the age of 12 or 13, in grade 6 or 7, they will start learning their second foreign language. Up until 2018, it was up to the schools to decide whether they wanted to introduce the second foreign language in grade 6 or 7 (Skolverket, 2016). As of 2018, however, it has to be introduced in grade 6. Most Swedish schools provide French, German and Spanish, but other language options exist, for example Chinese, Arabic or Italian. Multilingualism in Sweden has been investigated from different points of view (Bardel, Falk, & Lindqvist, 2013; Bijvoet & Fraurud, 2010; Boyd & Huss, 2001; Ganuza, 2008; Svensson, 2009). However, while much of the previous research regards the use and the learning of the Swedish language in multilingual settings, there is a research gap when it comes to the multilingualism of younger learners of French in Sweden and its didactic implications (but see Guijarro-Fuentes, Schmitz, & Müller, 2015 for the learning of French in other multilingual settings, and Lindqvist, 2009, 2010, 2012, for studies on adult Swedish learners’ learning of French as a third language). Thus, following the most recent curricula, formal learning of French in school is introduced in grade 6, which means that it begins around three years after the start of the formal learning of English. Thus, in this respect, French can be considered as a third language (L3) for most Swedish pupils in the chronological sense, being the third language they encounter in life (cf. Cenoz & Jessner, 2000). The multilingualism of Swedish pupils can be more complex than that, however, and this is reflected in the writing of the participants in the present study (see Sect. 4 for a description of the task and the participants). The use of multiple languages can be explained by the fact that in grade 8, one or two years after the introduction of the second foreign language, many pupils choose to start studying yet another foreign language, which means that they undergo learning processes in several languages simultaneously. The typical language repertoire for a Swedish pupil taking French from grade 6 and Spanish, for example, in grade 8, with Swedish as his or her only mother tongue, would thus be: mother tongue Swedish, then English, French and Spanish as foreign languages. The question then becomes delicate as to the labelling of these languages. Staying with the chronological definition proposed above, Swedish would be the L1, English the L2, French the L3 and Spanish the L4. However, within the field of L3 acquisition research, the more common definition used seems to be that of Hammarberg (2009), according to which the L3 is the language that is currently being learned. In our case, we would thus have three L3s (English, French and Spanish), as the pupils study them at the same time in school. In the present study, however, we will regard French as the L3 of the participants of the study in order to avoid terminological confusion. It is a language they are learning at the time of the data collection, and it is also the language under study from the researcher’s point of view. Coming back to the question of migration, nowadays it is more and more common for Swedish pupils to have knowledge of yet other languages spoken at home.

Didactic Challenges in the Multilingual Classroom—The Case …

89

Thus, it is clear that the learning of French by Swedish learners usually takes place under multilingual circumstances and therefore it can constantly be affected by other languages. The two reasons mentioned above, i.e. migration and the introduction of a second foreign language in school contribute to the widespread multilingualism in Sweden. The term “multilingualism” is here used in order to refer to knowledge of at least three languages, including the mother tongue(s). The proficiency level of each of the languages is not taken into account in the present study. In fact, many previous studies show that any background language [L1(s) and/or L2(s)] might be activated and used during L3 production, independently of proficiency level (De Angelis, 2005; Lindqvist & Bardel, 2013). The overall purpose of the present study is to examine and discuss the use of multiple languages in the school context of formal L3 learning. The use of multiple languages will be operationalized as cross-linguistic influence (CLI), that is, the use of the L1 and/or the L2(s), in 105 multilingual learners’ writing. The learners are pupils in grades 6, 7, 8 and 9 in a Swedish school. The paper takes the example of French, a language that is usually learned as a third language in Sweden, as discussed above. A further aim is to strengthen the link between research on language learning and language didactics. It is our conviction that language didactics would benefit from taking results from current research on various aspects of language learning into account to a larger extent (cf. Alonso Gonzalez & Rothman, 2017).

2 Background Research within multilingualism and third language acquisition has flourished in recent years (for a historical overview see Jessner, 2008). One of the main areas of research within these fields is the study of cross-linguistic influence. Several attempts have been made to explain why the background languages are activated during L3 production, and also why a particular background language in the learner’s language repertoire turns out to be the main source of influence. By now, it seems undisputable that several factors interact in the activation of the background languages, and that it is seldom possible to isolate one single determining factor (Lindqvist, 2015; Lindqvist & Bardel, 2013). Rather, there seems to be an intricate interplay between several factors, depending on, for example, the language constellations under study. One of the most cited factors within the L3 field seems to be the L2 status factor, which was originally proposed by Williams and Hammarberg (1998) in their longitudinal case study on L3 Swedish. The factor has since then been elaborated in various publications by Falk and Bardel (2011) or Falk and Bardel (2012). They take the point of departure that there are many cognitive and situational similarities between L3 learning and the learning of other foreign languages, as opposed to the acquisition of the L1. From this follows that previously learned foreign languages, rather than the L1, would be more likely to be activated in L3 learning. Various studies have found evidence for the strong role played by the L2 status factor (Falk, 2010; Sánchez, 2011b).

90

C. Lindqvist

Another explanatory factor for CLI is that of typology. One of the earliest and most cited accounts of the important role of this factor within the L3 field is Ringbom (1987), who showed that Swedish rather than Finnish tended to influence learners’ L3 English, this being the case both for learners with Swedish as L1 for learners with Swedish as L2. According to Ringbom, this result was due to the fact that Swedish and English are more closely related. Furthermore, Ringbom often stressed that learners constantly try to look for similarities between the languages they know (see for example Ringbom, 2007). Ringbom (2007) also pointed out that typology was particularly important in the beginning of the learning process. Many studies support the determining role of typology in L3 acquisition and use (Cenoz, 2001; De Angelis, 2005; De Angelis & Selinker, 2001; Lindqvist & Bardel, 2013; Odlin & Jarvis, 2004). The role of typology in L3 French has been investigated in a few studies, and studies on oral production (Lindqvist, 2009; Singleton, 1987) as well as written production (ÓLaoire & Singleton, 2009) have shown that a typologically related language often dominates as a source for CLI. In a recent study, examining CLI in written L3 French, Lindqvist (2015) showed that Swedish learners were influenced by L2 English to a larger extent than by L1 Swedish. English is more closely related to French than is Swedish, especially as far as vocabulary is concerned. It also turned out that the learners’ perceptions of the relatedness between Swedish, English and French, as investigated in a typological questionnaire, were decisive for the source of CLI. Thus, the learners resorted more to English, the language that they perceived as being closer to French in all respects, than to Swedish when writing in French L3. That said, previous studies on L3 French have also suggested that a more distant language (Swedish) can have a strong impact (Lindqvist, 2009). There is now an abundant body of research on multilingualism and L3 learning showing that previous languages might influence the L3 in various ways. This should have consequences for the teaching of third languages, which has also been noted elsewhere (see Alonso Gonzalez & Rothman, 2017). Jessner (2008) discusses some of the implications. According to her, there should be a clearer focus on the multilingual learner as well as the multilingual teacher within third language teaching. For example, she argues that “the use of previous foreign language learning experiences and strategies as well as the development of skills to compare, transfer and infer should be fostered in TLA” (Jessner, 2008, p. 39). However, she goes on to say that this is nevertheless rarely the case in the multilingual language classroom today, suggesting that “prior linguistic knowledge of students can be exploited and not regarded as some kind of negative, interfering and destructive force that hinders the language learning process”. Naturally, the multilingual classroom implies specific challenges for the teacher. Jessner (2008, p. 41) claims that the “ideal language teacher” is multilingual and a language learner herself, having consequently developed an awareness of the language learning process. Furthermore, Jessner et al. (2016) stress that a multilingual approach in the classroom enhances the pupils’ metalinguistic awareness and facilitates further language learning. In addition, Lindqvist and Falk (2017) argue that language learners need to develop their knowledge about the languages in the classroom. Even though it is an impossible task to master all the languages in the classroom, it seems reasonable to assume that teachers would

Didactic Challenges in the Multilingual Classroom—The Case …

91

benefit from having some basic knowledge about the grammar in the most common languages in the classroom. With this knowledge, it will be easier for the teacher to understand the pupils’ language use.

3 Research Questions The present study addresses the following research questions: 1. To what extent do multilingual learners use their different languages while writing in L3 French? 2. What languages [L1 and/or L2(s)] are used for lexical and grammatical crosslinguistic influence, respectively? 3. Is there a difference regarding the use of the background languages between grades?

4 Methodology 4.1 Participants The participants are 105 Swedish pupils in grades 6, 7, 8 and 9 in a school in the Stockholm area.1 The data were collected after having obtained the consent from both the teacher and the pupils’ parents. The pupils as well as the parents were assured that the data were gathered for research purposes only, and thus would not be available at all to their teacher, for example, for assessment purposes. Anonymity was also guaranteed, i.e. both parents and pupils were informed that the pupils’ identity would not be revealed. The pupils were asked to fill in a background questionnaire that mainly regarded their language knowledge. All the pupils had Swedish as their first language (see Table 1). They had all started learning English as their first foreign language in grade 1, 2 or 3, that is at the age of 7, 8 or 9, and had thus learned English for at least three years and up to eight years at the time of the data collection. Swedish school pupils generally have a relatively good knowledge of English from an early age, due to a high amount of input from various media and input from the surrounding society, usually on a daily basis. It is therefore important to stress that English is their L2, both from a chronological point of view, and probably also from a proficiency point of view, although this has not been tested. French is thus their second foreign language, or third language, L3. French has always been taught from grade 6 in this particular school. As mentioned above, Swedish schools could choose 1 Part of this data set has been analyzed in a previous study (Lindqvist, 2015). In that study, however,

only pupils with L1 Swedish, L2 English and L3 French were included due to the research questions asked.

92

C. Lindqvist

Table 1 Participants Grade

Age

Number of pupils

Semesters of study of French

L1

L2(s)

6

11/12

17

1

Swedish

English (17) Finnish (1) Norwegian (1) Polish (1)

7

12/13

26

3

Swedish

English (26) Danish (1)

8

13/14

35

5

Swedish

English (35) Chinese (2) Spanish (16) German (1)

9

14/15

27

7

Swedish

English (27) German (2) Spanish (12) Chinese (4)

to introduce the second foreign language in either the 6th or the 7th grade up until 2018 (Skolverket, 2016). The data collection took place in January/February, which means that the 6th graders had only been taught French for one semester, the 7th graders for three semesters, and so on. Importantly for the present study, from grade 8, it is possible, but not mandatory, for pupils in the Swedish school system to start learning an additional foreign language, Spanish or German, for example. That is why the number of different L2s increases among the pupils from grade 8 onwards. Table 1 shows the background information regarding the participants. All in all, knowledge of eight different foreign languages has been reported in the questionnaires, the most common ones being Spanish and German, apart from English.

4.2 Task and Data Collection The participants were asked to retell a short cartoon, The Dog Story, in writing (see Appendix). This particular short cartoon has been used in some previous L3 studies (e.g., Lindqvist, 2015; Sánchez, 2011a, 2011b). It contains six pictures telling the story about two children, a boy and a girl, who, waving goodbye to their mother, leave their house for a picnic. They have brought a picnic bag with them. As they arrive in the woods it turns out that their dog had been hiding in the basket and that he has eaten all the food. The pupils were invited to try to tell the story in writing in French, using only pencil and paper. They were given 20 minutes of an ordinary class for the task. They were not allowed to work together or use dictionaries or any other help. They were encouraged by the teacher and the researcher to try to write as much as they could and to try to communicate what they saw on the pictures. The

Didactic Challenges in the Multilingual Classroom—The Case … Table 2 Total number of words produced in the retellings

Grade

Words produced

93

Range

Average

6 (N = 17)

503

7–74

26

7 (N = 26)

1835

18–164

66

8 (N = 35)

3516

34–197

100

9 (N = 27)

3066

62–263

114

length of the writings turned out to vary considerably, probably due to the differences in proficiency between the grades. The lowest number of words produced was seven (for a grade 6 pupil) and the highest was 263 (for a grade 9 pupil). The total number of words as well as the average in the different grades are shown in Table 2. It is clear that the average number of words increases from year to year, with the highest average in year 9 and the lowest in year 6. However, the total number of words is highest in grade 8. This is explained by the fact that the number of pupils is also the highest in this grade. It can also be noted that the number of words produced has a considerable range in all grades.

4.3 Identification and Categorization of Cross-Linguistic Influence All the retellings were analyzed in search for traces of languages other than French. Specifically, two types of cross-linguistic influence were included in the analysis: lexical and grammatical CLI. These two types are described below.

4.3.1

Lexical CLI

There are two types of lexical CLI in the data. However, they will be accounted for as one common category of lexical CLI in the present study. The first type includes occurrences of pure code-switches, as seen in example (1), where an English word has been inserted into the otherwise French sentence. (1) Il est dans le basket. (P86, grade 8) ‘He is in the basket.’ Target word: panier.

The second type of lexical CLI includes instances of adaptations of words from one of the background languages into French. This procedure has resulted in deviant word forms in French, as in example (2), where, again, the word panier (‘basket’) seems to be unknown to the learner.

94

C. Lindqvist (2) Le chien est dans la basquette. (P113, grade 9) ‘The dog is in the basket.’ Target word: panier.

4.3.2

Grammatical CLI

Grammatical CLI are instances of deviant word order, genitive constructions, lack of articles and adaptations of the progressive into French that can be traced back to one of the background languages. In (3), the placement of the adverb ‘aussi’ seems to follow English word order (in Swedish, the corresponding word ‘också’ would normally follow the French word order and it could also be placed after the verb, but not before.) (3) Je aussi vois une chemise. (P162, grade 6) ‘I also see a shirt.’ Target word order: Je vois une chemise aussi.

One example of a lack of the definite article is shown in example 4. (4) Le garçon aime fromage, mais il aime fraise aussi. (P13, grade 7) ‘The boy likes cheese, but he also likes strawberry.’ Target construction: Le garçon aime le fromage, mais il aime les fraises aussi.

In French, the definite article is obligatory in these cases. All instances of lexical and grammatical CLI were identified in the data and categorized according to the source of the CLI, L1 or L2. However, as is often the case in studies including multilingual data, one category with cases where it is impossible to determine the source language had to be included (see discussion in Lindqvist & Falk 2014). This is the case in the following example (5), where the verb ‘être’ is used when expressing age. In French, the verb ‘avoir’ should be used here. The choice of ‘être’ could be traced back to either Swedish or English.2 (5) Ils sont cinq ans. (P55, grade 8) ‘They are five years old.’ Target: Ils ont cinq ans.

2 Similarly,

instances of adaptations of the progressive form could be due to the existence of that structure in both English and Spanish. However, none of the participants who produced that form had knowledge of Spanish. That type of influence will thus be categorized as CLI from L2 English.

Didactic Challenges in the Multilingual Classroom—The Case …

95

Table 3 Results lexical CLI Grade

L1 Swedish

L2 English

L2 Spanish

L1 Swedish/L2 English

6 (N = 17)

6

17

1a

0

24

4.77

7 (N = 26)

20

56

0

5

81

4.41

8 (N = 35)

62

98

5

21

186

5.29

140

4.56

431 (100%)

9 (N = 27)

9

102

17

12b

Total (N = 105)

97 (23%)

273 (63%)

23 (5%)

38 (9%)

Total CLI

Proportion CLI/number of words (%)

a Although

this particular learner had reported that his or her only L2 was English, and that his or her only L1 was Swedish, one instance of a Spanish word was found in the data: madre (‘mother’) b There was one instance that could be of either Spanish or Swedish origin: sol (target word soleil). Moreover, there were three occurrences of the word park, which could be code-switches to either Swedish, English or German

5 Results 5.1 Lexical CLI The results clearly show that the pupils are influenced by their background languages to a relatively large extent. In particular, there are four things to note from Table 3. First, lexical CLI predominantly comes from the L2s. Leaving the mixed category aside, CLI from the L2s represents 75% of the occurrences. Secondly, among the L2s known to the learners, English is the language that dominates in lexical CLI. Thirdly, Spanish is the only L2 that is used apart from English. Recall that the pupils have knowledge of eight different L2s in total. However, Spanish is also the additional L2 that most learners know (Table 1). Finally, the fourth thing to note is that even though the number of instances of lexical CLI is considerably higher in grades 8 and 9, the difference between the grades does not seem significant when looking at the proportions in relation to the number of words produced. These results will be discussed in Sect. 6.

5.2 Grammatical CLI In comparison to lexical CLI, the number of occurrences of grammatical CLI is lower (Table 4). The results pertaining to grammatical CLI show that the mixed category L1 Swedish/L2 English dominates. This is of course due to the many grammatical similarities between these languages, which make a separation between them impossible in an analysis such as the present one (Lindqvist & Falk, 2014). Where

96

C. Lindqvist

Table 4 Results grammatical CLI Grade

L1 Swedish

L2 English

L1 Swedish/L2 English

Total

Proportion CLI/number of words (%)

6 (N = 17)

0

1

3

4

0.79

7 (N = 26)

1

2

20

23

1.25

8 (N = 35)

2

11

22

35

0.99

9 (N = 27)

3

2a

13

18

0.58

Total (N = 105)

6 (8%)

16 (20%)

58 (73%)

80 (100%)

a One

of these instances was produced by a learner with knowledge in Spanish. It might thus be the case that the influence comes from Spanish or English in this case. However, none of the other pupils with Spanish as L2 produced similar forms

such a distinction is possible, however, it seems that English is slightly more often the source of influence. It is interesting to note that no other L2 seems to be influencing French at the grammatical level.

6 Discussion The first research question was: To what extent do multilingual learners use their different languages while writing in L3 French? The results show that, overall, they tend to use many of their languages in this situation, and to a relatively large extent. There are 431 instances of lexical CLI and 80 instances of grammatical CLI, that is, 511 occurrences in total. This means that each pupil produces 5 occurrences of CLI on average in their writings. However, even though there are quite a few instances of CLI, and despite the fact that the pupils have knowledge of several languages, only three of them dominate: Swedish, English and Spanish (Tables 3 and 4). Furthermore, the results show that lexical CLI is far more frequent than grammatical CLI overall. It is thus clear that in our data, the languages pupils resort to are languages they have learned, or rather, are learning, in school, i.e. other foreign languages. In other words, they do not seem to activate languages spoken at home. This result could be explained by the L2 status factor (Falk & Bardel, 2012). As we saw in the background section, in their model, Falk and Bardel (2012) stress the fact that languages that are learned formally under the same circumstances have a tendency to influence one another. This would explain why English and Spanish are frequently used in the French retellings. However, the pupils do not make use of the other formally learned L2, German. Thus, even though the L2 status factor does seem to come into play, it is not the only explanatory factor. Why do pupils use Spanish and not German? The obvious explanation seems to be found in the typology factor, Spanish being more closely

Didactic Challenges in the Multilingual Classroom—The Case …

97

related to French than German is. It is plausible that the learners take advantage of that, as in the study by Lindqvist (2015). The second research question was: What languages [L1 and/or L2(s)] do they use for lexical and grammatical cross-linguistic influence, respectively? The results show that as for lexical CLI, English clearly dominates as the source language with 63% of all occurrences. Moreover, it is potentially involved in 9% of the cases for which the source is not possible to isolate. Swedish is the second most used language for lexical CLI with 23% of the occurrences. It is interesting to note that among all the other L2s known to the learners, Spanish is the only language that they use for lexical CLI. English is also more used than Swedish in grammatical CLI (20 vs. 8% of the occurrences), although most of the instances are traceable to both Swedish and English (73%). No other language seems to be the source of grammatical CLI. The fact that English is the only L2 that seems to be activated could be explained by the pupils’ relatively high proficiency in this language, as opposed to the other L2s. Clearly, a certain grammatical structure must have been acquired in order for it to be transferred. Recall that the pupils have only studied Spanish and German for one (8th graders) or three (9th graders) semesters. As mentioned in Sect. 4, these learners have studied English for at least 5 or 6 years and generally have a high exposure to the language, so even though their proficiency level has not been assessed it is very likely that they are considerably more proficient in English than in their other L2s. The third research question was: Is there a difference regarding the use of the background languages between grades? It can be concluded that there are both differences and similarities. Starting with the similarities, the results show that the proportion of both lexical and grammatical CLI is practically the same over the grades. This is an interesting finding, since many previous studies have shown that cross-linguistic influence tends to be particularly frequent in the beginning of the learning process (Hammarberg, 2009; Lindqvist, 2009). We can only speculate about the reasons for this finding, but it might be the case that the difference in proficiency level between the grades is not that pronounced, at least not big enough to reveal differences in this respect. Earlier studies have seen this difference when comparing learners who are clearly at different proficiency levels. Lindqvist (2009), for example, compared beginning learners with intermediate and advanced learners and the results clearly showed a decrease in the proportion of CLI. However, the learners in the present study can probably not be that easily separated into different proficiency levels because they have studied French for a relatively short period of time. Although the average number of words produced in the retellings seems to indicate an increasing proficiency, which could indicate more advanced writing and a more expanded vocabulary, the differences in proficiency is probably not that big. There are also differences between the grades, the most obvious and perhaps expected one being that those pupils who have started learning additional foreign languages make use of them in their French writing. However, Spanish is the only L2 used apart from English for lexical CLI. As far as we can tell, there are no traces from other languages. This finding can be explained by the relative closeness between Spanish and French, especially in comparison with other L2s.

98

C. Lindqvist

7 Pedagogical Implications There are several didactic implications from the results of this study. As discussed in the background section, the teaching of third languages in multilingual classrooms would benefit from integrating more research findings from the fields of multilingualism and third language learning (Alonso Gonzalez & Rothman, 2017). The present study adds to the results of previous research by showing that multilingual learners’ writing manifests cross-linguistic influence from several languages. It has repeatedly been shown in L3 studies that other foreign languages are activated in L3 learning and use, and the present study is no exception. Furthermore, cross-linguistic influence is manifested from three different languages in our study, even from those in which the learners have a relatively low proficiency. Accordingly, the multilingualism of the learners should be taken into account in the language classroom. As we have seen, Jessner (2008) argues that the ideal third language teacher should have the experience of language learning and should also be multilingual herself. We agree that these competencies would increase the teacher’s awareness of the language learning process and consequently the possibilities of understanding the pupils’ language use. On a more practical note, in their feedback on written tasks and also when giving oral feedback in the classroom, teachers with some basic understanding of the pupils’ languages should be able to point to instances of CLI rather than mark them as errors. That would create a learning situation and would also contribute to the metalinguistic awareness of the pupil, who would be given the possibility to reflect upon differences and similarities between the languages he or she knows (Jessner, Allgaüer-Hackl, & Hofer, 2016). As Ringbom (2007) often argued, it is a natural process for learners to constantly seek for similarities between languages, and, as we have seen, this is manifested in their language use. Even though it might not always be possible for the teacher to master all the languages in the language classroom, explicit metalinguistic knowledge about the languages can aid (Lindqvist & Falk, 2017).

8 Conclusion The present study has shown that Swedish pupils resort to their background languages relatively frequently when writing in French. Cross-linguistic influence occurs both at the lexical and the grammatical level, although lexical CLI dominates in the data. Interestingly, the proportion of CLI is roughly the same over the years, which indicates that the background languages are present even as pupils develop their knowledge of the French language. In fact, as other L2s are introduced in their schedules, these languages also seem to be activated and available for use in the L3. This is the case with Spanish among the 8th and 9th graders, but English too seems to be constantly activated among the pupils, from grade 6 onwards. The results clearly show that both Swedish and English influence the pupils’ writing in French, with respect to both vocabulary and grammar.

Didactic Challenges in the Multilingual Classroom—The Case …

99

The multilingual classroom is a reality and we need to take this into account in third language teaching. Language teachers today and in the future should be prepared for this situation. It thus becomes evident that more focus should be put on CLI and multilingualism within the language teacher programs in order to increase future teachers’ awareness of multilingual learners’ language learning and use. The present paper has discussed this situation from the point of view of L3 French in Sweden, but we are confident that it is by no means a rare case. Rather, it pertains to various language learning situations in many parts of the world.

Appendix: The Dog Story

Syntactic Transfer in L3 Learning. What Do Models and Results Tell Us About Learning and Teaching a Third Language? Camilla Bardel

Abstract In this chapter, five theoretical models of syntactic transfer in third language (L3) learning are presented together with results from studies that examine the role of the background languages (L1 and L2) in L3 syntax. The models are the Cumulative Enhancement Model (CEM, Flynn, Foley, & Vinnitskaya 2004), the L2 status factor hypothesis (Bardel & Falk, 2007, 2012), the Typological Primacy Model (TPM, Rothman, 2011, 2015), the scalpel model (Slabakova, 2017), and the Linguistic Proximity Model (LPM, Westergaard, Mitrofanova, & Mykhaylyk, 2017). With these models, L3 syntax has recently and quickly become a debated issue in the field of Second Language Acquisition (SLA). The models deal with various factors that are held to play particularly important roles in the initial stages of L3 learning. The factors are, above all, Universal Grammar, typological relations between languages or between structures, the learner’s perception of similarities between languages, and the level of metalinguistic knowledge and proficiency in the involved languages. Empirical data tend to point in different directions regarding the significance of these factors. The overall results point at the dynamic nature of multilingualism in that they indicate that both the L1 and the L2(s) may act as transfer sources in L3 syntax, but questions concerning which factors lead to transfer from which background language, and of which particular structures, remain unsolved. This chapter surveys the five models and their attempts to answer the question of how previously acquired or learned languages play a role in the learning of L3 syntax. It ends with a discussion of what this line of research can offer language teachers. Keywords L3 learning · L3 syntax · Multilingual transfer · Metalinguistic knowledge · Pedagogical implications of L3 research

C. Bardel (B) Institutionen för Språkdidaktik, Stockholms Universitet, 106 91 Stockholm, Sweden e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_6

101

102

C. Bardel

1 Introduction Multilingual contexts are common all over the world. As a consequence of, and a prerequisite for migration, new lifestyles and new ways of communication, multilingualism can be expected to expand (Aronin, Fishman, Singleton, & O’Laoire, 2013). It will therefore be increasingly important for language educators to take into consideration that language learners often have multilingual backgrounds, and that this will have implications for their additional language learning in terms of multiple available linguistic resources and potential gains related to raise metalinguistic awareness, assumed to be an effect of multilingualism (Cenoz, 2009; Jessner, 2008). The interaction of the languages known to the multilingual speaker is a phenomenon that has been shown to play a role in additional language development (Aronin et al., 2013; Auer & Wei, 2007; De Angelis, 2007; Hammarberg, 2009, 2010). This chapter addresses the question of how previously acquired or learned languages—first, second or foreign languages (L1/L2s/FLs)—interact with the syntax of a third or additional language (L3/Ln). The main aim of the chapter is to discuss some of the attempts to answer this question that have been put forward in previous research. A more specific but secondary aim is to examine what research into L3 syntax can offer language teachers. Five existing models of L3 syntax will be highlighted and their relevance and implications for language education will be discussed. The models are the Cumulative Enhancement Model (CEM) (Flynn, Foley, & Vinnitskaya, 2004), the L2 Status Factor Hypothesis (L2SFH) (Bardel & Falk, 2007, 2012), the Typological Primacy Model (TPM) (Rothman, 2011, 2015), the scalpel model (Slabakova, 2017), and the Linguistic Proximity Model (LPM) (Westergaard, Mitrofanova, & Mykhaylyk, 2017). As will be shown, empirical results from studies on L3 syntax that support the different models indicate that both the L1 and the L2 may have an impact on the process of L3 learning, in the sense that structures from all acquired or learned languages can be transferred into an additional language. The theoretical underpinnings of the above-mentioned L3 models, as well as the empirical data from the studies they build on, could be of interest to those engaged in third language teaching and learning, that is, teachers, teacher students, teacher educators, and learners themselves. In particular, metalinguistic knowledge and learners’ capacity of noticing of similarities and differences between languages are issues that will be considered when discussing the implications of the models for language education. The chapter will start with a section on the notions of multilingualism, transfer, and cross-linguistic influence. The concept of transfer has a longstanding history in Second Language Acquisition (SLA) research and serves as a background to current views on L3 learning. But first, the notions of formal language learning and informal language acquisition will be addressed briefly. Although the distinction between the two is not always clear-cut, it is relevant for the discussion of some of the reviewed transfer models and for the idea of metalinguistic knowledge playing a role for transfer into an L3. Formal language learning is often defined in relation to its counterpart informal language acquisition (Paradis, 2004, 2009). Following Paradis, this distinction will be adhered to when referring to different language learning settings (except

Syntactic Transfer in L3 Learning. What Do Models and Results …

103

when citing authors who do not make this distinction). While formal language learning is typically—but not categorically—the case when it comes to the study of foreign languages and the language classroom, in, or outside of, the community where the TL is used, informal language acquisition happens when children acquire their L1 in a natural setting.1 Also, older learners may acquire an L2 informally if the second language is appropriated in a community where the TL is used by native speakers. It is important to remark, though, that as soon as a second language becomes a classroom subject, there are formal learning aspects coming into play, even if the setting is in the community where the TL is spoken. It follows that also in the case of literacy development in the L1, there is an aspect of formal learning (Falk, Lindqvist, & Bardel, 2015). Hence, rather than representing a true dichotomy, the two categories formal language learning and informal language acquisition appear to be situated on a continuum.

2 Multilingualism, Transfer and Cross-Linguistic Influence A strong line of current SLA research considers humans as potentially multilingual by nature, and multilingualism as the normal state of linguistic competence (Hammarberg, 2010). As a matter of fact, a great part of the world’s population is bi- or multilingual (Hammarberg, 2009; Mackey, 1967; Tucker, 1998). Multilingualism is today used by many researchers as the overarching term for speakers who have knowledge of more than one language, independently of proficiency level. Furthermore, language learning is generally acknowledged as a dynamic process: the character, proficiency level and modality of use of all the languages known to the individual speaker may vary not only synchronically, but typically also change over time. As De Bot and Jaensch (2015, p. 140) put it, multilingualism is a process, not a state, and languages are continually changed and reorganized, at the individual as well as the social level. When it comes to the addition of new languages, the fact that many language learners are already speakers of one or several non-native languages should not be neglected (De Angelis, 2007) and learners’ multilingual backgrounds have increasingly come into focus with the growing interest in multilingualism and multilingual education (Hammarberg, 2018). Reviewing a number of studies in her book, mainly within the lexical area, but also some on L3 syntax, De Angelis (2007, p. 4) points out that “a general theory of non-native language acquisition cannot be based on L2 learner behaviour alone”. As suggested by De Angelis (2007) and Hammarberg (2009) inter alia, learning a L3 is different in many respects from learning a L2. First of all, it differs from chronologically true L2 acquisition, that is, the first new language learned directly after the L1, simply because more background languages 1 Another

distinction is the one between foreign language and second language, normally used to distinguish between languages that are studied formally versus acquired informally, in an environment where other languages are used for communication versus in the community where the TL is used for communication. We will not go deeper into this distinction in this chapter.

104

C. Bardel

may come into play. Furthermore, according to the viewpoint taken in this chapter, it is important to consider that L3 differs from L2 learning essentially because the L3 learner already has knowledge of at least one non-native language, that is, one or several L2s (Bardel & Falk, 2007, 2012; Bardel & Sánchez, 2017; Falk & Bardel, 2010, 2011). From this perspective, the age factor plays an important role in L3 learning. The age of onset is a major factor that distinguishes L2 and L3 learning from L1 acquisition, when both the L2 and the L3 are learned later than the L1 (Bardel & Falk, 2012; Hammarberg, 2016, 2018). Cases of simultaneous bilingualism (2L1), followed by later L2 and L3 learning, present a different picture, as will be seen later in this chapter. An important research development has lead towards the insight that bilinguals or multilinguals are not to be seen as individuals comparable with two native speakers of two different languages, but rather as speakers of one dynamic system or a system with different interacting subsystems (De Bot & Jaensch, 2015; Paradis, 2004, 2009). In the multilingual brain, “all linguistic knowledge is interconnected and the different languages of an individual are not functionally separated” (Slabakova, 2017, p. 653). The dynamic multilingual system is interactive and there is potential influence at all linguistic levels (lexicon, syntax, phonology etc.) between all languages known to some extent by the individual. Hence, languages do not influence each other exclusively in the order that they have been acquired or learned, but all languages present in the multilingual mind may interact. When a multilingual individual encounters a new language, the L1 and all subsequently appropriated languages have the potential to influence the new language, and the most recently encountered language can also influence those acquired or learned earlier. Bi- or multidirectionality is characteristic for cross-linguistic interaction between the languages of the multilingual mind (Cook, 2003; Herdina & Jessner, 2002; Kroll & Dussias, 2017). Although multidirectionality is not the focus of this chapter, its presence should not be overlooked. The notion of transfer has a longstanding history (Jarvis 2017), and while there have been proposals of alternative terms, ‘transfer’ is still used generally in the field of SLA and in the area of L3 study. With the Contrastive Analysis Hypothesis (CAH), Fries (1945) and Lado (1957) suggested that problems with learning specific linguistic patterns could be predicted by comparing the learner’s L1 with the language to be learned (the target language, TL) and looking for differences. Interference—a term that was used by Weinreich (1953) to denote non-target like structures (errors) that reflected the learner’s native language—soon acquired negative connotations and more or less fell out of use. The idea of predictable transfer from the L1 to the L2 has been elaborated and refined in a number of studies from the middle of the 20th century and onwards (Anderson, 1983; Golden, Jarvis, & Tenfjord, 2017; Jarvis, 2016; Jarvis & Pavlenko, 2008; Kellerman, 1983; Odlin, 1989; Odlin & Jarvis, 2004; Selinker, 1972), extending it to multilingual systems and to diverse levels of linguistic analysis, for instance to lexical semantics and to the discursive functions of language. Odlin (1989, p. 27) defines linguistic transfer as “the influence resulting from similarities and differences between the target language and any other language that has been previously (and perhaps imperfectly) acquired”. As suggested by De

Syntactic Transfer in L3 Learning. What Do Models and Results …

105

Angelis and Selinker (2001), native-language transfer and interlanguage transfer can be distinguished as two different types of transfer. Transfer can be positive or negative, positive transfer occurring when the learner assumes similarities that are compatible with objective similarities between the involved languages, and negative transfer occurring when the assumed similarities are in conflict with objective differences between languages (Jarvis & Pavlenko, 2008, p. 182). Furthermore, the term cross-linguistic influence (Ringbom, 1986, 1987; Sharwood Smith & Kellerman, 1986) has gained ground as an alternative to transfer, perhaps because of its broader scope. Cross-linguistic influence (CLI) refers to the general phenomenon of the learner’s relying on all previously acquired or learned languages, when encountering a new one. Originally, CLI was assumed to include instances of native-language transfer, interlanguage transfer, avoidance due to influence of another language, and even transfer from an interlanguage to the native language (Sharwood Smith & Kellerman, 1986, and above; for further discussion, see Jarvis, 2017). When it comes to multilingual learners, transfer into the TL can originate from the L1 or from an L2, and therefore in L3 learning, we may talk about multilingual transfer (Rothman, 2015). The huge body of studies on syntactic transfer from the L1 to the L2 (some are reviewed below) constitutes an important basis for L3 research; however, it most often disregards the potential role of other languages in the learners’ repertoire. In the field of generative linguistics, the role of transfer from the L1 in L2 syntactic development came to be seen as evidence for or against certain standpoints in the debate on the L2 learner’s (full/partial/no) access to Universal Grammar (UG). According to the generative paradigm, UG is assumed to be present in all L1 speakers’ minds, and to be a necessary prerequisite for child L1 development (White, 1989), but the ideas differ when it comes to the role of UG in non-native language development. In the 1990s, a series of studies led to a number of hypotheses about the development of L2 syntax, and about which types of linguistic structures, or which syntactic features, learners transfer from the L1 to the L2 (Epstein, Flynn, & Martohardjono, 1996; Hawkins & Chan, 1997; Schwartz & Sprouse, 1996; Vainikka & Young-Scholten, 1996). These hypotheses differed as to how much impact the L1 grammar was supposed to have on the L2, and at what stage of L2 development L1 transfer was expected to occur. Schwartz and Sprouse argued in favour of full transfer at the initial state of L2 acquisition. According to the Full Transfer/Full Access Hypothesis (FT/FA), all syntactic properties of the L1 initially constitute a base for the developing L2 grammar, which is constructed with access to UG. The other transfer hypotheses did not predict total transfer of the L1 grammar, but rather different levels of involvement of the L1 grammar. For instance, Vainikka and Young-Scholten (1996) suggested that only lexical categories transfer from the L1. Epstein et al. (1996) proposed that L2 learners, although having access to UG, do not necessarily start out from their L1 when learning an L2. Another hypothesis, The Failed Functional Features (FFH) proposal by Hawkins and Chan (1997), assumed full transfer of L1 in the L2 initial state. Differently from the FT/FA however, according to which UG is fully available, the FFH hypothesized that UG is only partially available to L2 learners, and that parameters cannot be reset.

106

C. Bardel

The basic questions about syntactic transfer in non-native language learning feature also in L3 research, but the idea of L1 being the only source of influence in L3 syntax has been challenged by Bardel and Falk (2007), Bohnacker (2006), Flynn et al. (2004), Leung (2005) and Rothman and Cabrelli Amaro (2010), and in current L3 research (Slabakova, 2017; Westergaard et al., 2017), syntactic transfer from the L2, alongside the L1, is acknowledged (see below). Full transfer from either L1 or L2, depending on the overall similarity of the involved languages, has been suggested by Rothman (2011, 2015), and subsequently it has been suggested that both developmental conditions and differences in linguistic input as well as the character of particular linguistic structures determine transfer from the background languages (Slabakova, 2017; Westergaard et al., 2017). Returning to the terminological discussion above, the two terms transfer and cross-linguistic influence are today often used interchangeably (Slabakova, 2017; Westergaard et al., 2017), although some place important epistemological weight on the distinction: Transfer, as we intend the construct, refers to the level of grammatical mental representation for a given property, whereas CLI could be a simple bleeding over in performance and/or processing—influence in the truest sense of the word—from outside the interlanguage system. (Gonzalez Alonso & Rothman, 2017, p. 686)

The early SLA hypotheses seem to regard transfer from the L1 as a conscious process, assuming there to be a certain degree of metalinguistic awareness on the part of the learner. For instance, Anderson’s (1983) Transfer to somewhere principle was supposed to apply in cases where the learner perceived some kind of similarity between the languages in question. In the same year, Kellerman (1983) introduced the notion of psychotypology referring to the learner’s individual perception of the typological properties of the TL and its typological relation to other languages known by the learner, especially the L1. The interest for psychotypology has had a revival in recent L3 studies and is a topic of debate in studies by Hammarberg (2016), Rothman (2011), Rothman and Cabrelli Amaro (2010) and Williams and Hammarberg (1998). Languages that are closely related, such as Romance languages, share many features both at a lexical level and syntactic level. However, as Sánchez and Bardel (2017, p. 243) noticed, similarities between closely related languages “are not an all-ornothing question”, and learners may extend or overgeneralize similarities that exist between two particular languages to structures that differ. We will return to the role of awareness and consciousness in Sect. 3, in particular when discussing the L2SFH (Bardel & Falk, 2007, 2012; Falk & Bardel, 2010, 2011) and the TPM (Rothman, 2011, 2015).

3 L3 Syntax Models Lately, a number of studies have shown that not only the L1, but also the L2, can play a significant role for transfer of syntactic structures into the L3. Syntactic transfer from either the L1 or the L2, or both, into a L3 has developed into a special area of

Syntactic Transfer in L3 Learning. What Do Models and Results …

107

interest. In particular, studies conducted with quantitative methods have shown that closely related languages such as Romance languages affect each other in terms of syntactic transfer from the L2 to the L3 (Bardel, 2006, regarding the placement of the negation in Italian L3; Rothman & Cabrelli Amaro, 2010, regarding the pro-drop parameter in French and Italian as L3). These studies show that the L2 may take on a stronger role than the L1 as the main transfer source. This has been accounted for in terms of the L2 status factor (Williams & Hammarberg, 1998), which emphasizes the cognitive similarity between the L2 and the L3, both being non-native languages, as opposed to the L1 (Bardel & Falk, 2012; Falk & Bardel 2007, 2011), a hypothesis that has been further investigated by Sánchez (2015) and Neuser (2017). However, as pointed out by Rothman and Cabrelli Amaro (2010), a close typological relationship between the involved non-native languages may play a more important role than the fact that they are both non-native, that is, their L2 status. The disentangling of the L2 status factor and the role of typology could only be successfully achieved in studies of specific language combinations. In fact, Rothman’s subsequent studies, which combine L1s and L2s that vary with regard to how closely related they are to the TL (L1 English/L2 Romance/L3 Romance vs. L1 Romance/L2 English/L3 Romance), lend strong support to the idea that typology may trump the L2 status factor (see Sect. 4). Several models of L3 syntactic transfer have appeared in the literature during the last decades. Here, the focus will first be on three early suggestions, which have also been developed to varying extents in subsequent studies and are without doubt the most debated models up to now in research into L3 syntax. These are the CEM (Flynn et al., 2004), the L2SFH (Bardel & Falk, 2007, 2012) and the TPM (Rothman, 2011, 2015). As pointed out by Tavakol and Jabbari (2014, p. 117), these three models share a common feature: they take the position that L3 learning is not just another case of L2 learning, because it draws not only on the L1, but also on the L2, something which was ground-breaking when Flynn et al. (2004) published their study. These models, which have also previously been reviewed by García Mayo and Rothman (2012), García Mayo and González Alonso (2015) and Slabakova (2017) inter alia, all take similarity of some kind to be crucial for learning languages subsequently to the L1: general typological proximity at a global level (language relatedness), typological similarity between languages (limited to structures), and cognitive similarity (related to the learning process). The more recent models proposed by Slabakova (2017) and Westergaard et al. (2017) will also be examined here. They both take up the role of typology for transfer, and both claim to say more about stages beyond the initial ones than previous models, predicting that transfer does not occur in total but property by property, and that different structures transfer to different extents and at different points of development. They also posit that transfer can be detrimental (negative) as well as facilitative (positive), which is something they have in common with the L2SFH and the TPM.

108

C. Bardel

3.1 The Cumulative Enhancement Model The CEM builds upon the results from a series of studies on relative clauses in L2 and L3 (Flynn, 1983, 1989, reported on and followed up in Flynn et al., 2004). Flynn et al. (2004) is a pioneering study on L3 syntax, in the sense that it was one of the first that made predictions from different transfer hypotheses and showed multilingual syntactic transfer on the L3 target with a quantitative method, demonstrating empirically that L1 is not the only source for syntactic transfer in L3. The authors compared L3 learners of English (L1 speakers of Kazakh with Russian L2) with L2 learners of English (L1 speakers of Japanese). Kazakh and Japanese are identical in one syntactic setting (head-final and left-branching) whilst Russian and English are identical in another (head-initial and right-branching), which is relevant for the surface-structural differences as regards the relativisation of the noun phrase object within the subordinate relative clause structure in the involved languages. For example, the English (1) and the Japanese (2) structures in question would be: (1) John read [head the book [complement that Mary wrote]]. (2) John-wa [complement Mary-ga kaita [headhon-o]] yonda. John-theme Mary-nom wrote book-acc read ‘John read the book that Mary wrote’ (after Saito (1985) Flynn et al. (2004, p. 7)).

Results showed that the learners of English as L3 who had Kazakh L1 and Russian L2 learned the English Complementizer Phrase (CP) with more ease than Japanese L1 speakers who learned English as an L2. The authors concluded, specifically, that the L2 Russian facilitated the L3 learning, and more generally that language learning is cumulative. In a comparison of old L2 data and new L3 data, Flynn et al. (2004) clearly showed the existence of positive influence from similar structures in L1 to L2 and from similar structures in L2 to L3. The authors concluded that all languages known to the L3 learner can influence the development of subsequent learning, independently of whether they are an L1 or an L2, advocating a facilitative effect of any previously acquired/learned language (L1 or L2); where appropriate, any other language that is known either remains neutral or can enhance subsequent language acquisition. As posited by the authors themselves (Flynn et al., 2004, p. 5), this claim contrasts with models that either implicitly or explicitly characterize subsequent language learning fundamentally in terms of a deficit model (i.e., negative transfer).

3.2 The L2 Status Factor Hypothesis By the time that the L2SFH for syntactic transfer was first suggested as a model (Falk & Bardel, 2010, 2011), a number of studies had recently showed that the L2 may play a more significant role than the L1 as a transfer source in the L3 syntax (Bardel &

Syntactic Transfer in L3 Learning. What Do Models and Results …

109

Falk, 2007; Bohnacker, 2006; Leung, 2005). Results from these studies showed that L3 learners had a higher acceptance and production rate of structures that mirrored the L2 word order, than of structures that reflected the syntax of the L1. The notion of L2 status had originally been suggested by Williams and Hammarberg (1998) in their case study of Sarah Williams, a native speaker of English, who was learning Swedish as a L3. The L2 status was put forward as an important factor behind the L3 speaker’s choice of German L2 as a ‘supplier language’ in oral production of L3 Swedish, lying behind code-switches to German. It was defined as a general tendency to activate an L2 rather than the L1 when speaking in the L3, or as “a desire to suppress L1 as being ‘non-foreign’ and to rely rather on an orientation towards a prior L2 as a strategy to approach the L3” (Hammarberg, 2001, pp. 36–37). The following example from their study illustrates how Sarah asks how to say I find or I think in Swedish, using both German and English vocabulary, but clearly high-lighting the German word forms. (4) do the swedes say ‘f-finde ich’ or (Hammarberg (2009, p. 51) after Williams & Hammarberg (1998))

Bardel and Falk (2007) suggested that the L2 status factor may apply not only to the lexicon, but also to syntax in a L3, leading to both correct and incorrect word order in the L3. They conducted a study of negation and its relation to the finite verb in Swedish and Dutch as L3, both languages with post-verbal placement of negation in the main clause. They found that from two learner groups, one with post-verbal negation in their L1 and the other with post-verbal negation in their L2, the latter learned the TL structure more quickly than the former. In subsequent work (Bardel & Falk, 2012; Falk & Bardel, 2010, 2011), the L2 status factor was related to a neurolinguistic framework, namely the Declarative/Procedural model, henceforth the D/P model (Paradis, 2004, 2009, see also Ullman, 2015, 2016 for a similar account). According to this framework, there are two different, co-existing memory systems in the human brain, declarative memory and procedural memory. The declarative memory system is traditionally defined as the one that underlies explicit knowledge, that is, the kind of knowledge that can be brought to conscious awareness (Ullman & Lovelett, 2018, pp. 40–41). It is assumed by Paradis (2009) that knowledge of non-native languages that are learned in formal contexts tends to be explicit, although in certain circumstances already drawn attention to in the introduction, for instance if a language course is taken in the country where the TL is spoken, implicit linguistic competence may be built up in parallel. Being basically stored in declarative memory, non-native languages are more likely to display transfer phenomena from each other than from the L1 (Paradis, 2008, p. 344). This is likely to happen most often in the initial stages of L3 development, since the declarative system often operates initially, while the procedural system acquires analogous competence gradually, in parallel (Ullman & Lovelett, 2018, p. 42). Explicit metalinguistic knowledge is not unique for non-native languages, however. Rules and properties of the grammar of the L1 can also be learned explicitly, and as shown in a study on adjective placement by Falk et al. (2015), if

110

C. Bardel

learners have developed metalinguistic knowledge about how their L1 works, this may lead to transfer from the L1 into a non-native language in the initial state. A group of Swedish learners of Dutch L3 participated in the study, which tested whether learners, who had a high level of metalinguistic knowledge in the L1, exhibited transfer from the L1 to a relatively high degree. Differences in explicit metalinguistic knowledge of Swedish morphosyntax were controlled for by a metalinguistic test. Results showed that the learners with low levels of metalinguistic knowledge in L1 did not transfer from the L1, but rather from their L2. Learners with high levels of metalinguistic knowledge in the L1 transferred from their L1 to a higher extent. Because the structures in the L1 and the L3 coincided, this L1 transfer turned out to be facilitative. Falk et al. (2015) therefore argued that explicit metalinguistic knowledge, obtained by formal learning, transfers to additional languages that are formally learned as well, and that this mechanism lies behind the L2 status factor, but may also apply to explicit metalinguistic knowledge about the L1.

3.3 The Typological Primacy Model The TPM, suggested in Rothman (2010) and further elaborated in Rothman (2011, 2015), builds on the notion that (psycho-) typology (Kellerman, 1983; Williams & Hammarberg, 1998) and similarity between the background language (either L1 or L2) and the TL are crucial for transfer in L3 learning (for the notion of similarity, see Ringbom, 1987, 2007). Put simply, the basic claim of the TPM is that the background language (L1 or L2) which is most similar on the whole to the TL, will influence the representation and the processing of an L3, especially in the initial state. Similarity at diverse linguistic levels (syntax, phonology, lexicon, etc.) is common in closely related languages, although, needless to say, similarities also exist between less closely related languages, when it comes to structures, pronunciation and words. Following, to a certain extent, Schwartz and Sprouse’s (1996) Full Transfer/Full Access Hypothesis for L2 acquisition, the TPM predicts that one of the two systems (either L1 or L2) will be transferred completely in the initial stages of L3 development (i.e., full transfer). A number of studies on L3 learning involving English, Spanish and other Romance languages in different combinations have lent support to this model. For example, in Rothman (2010), two syntactic phenomena were tested in learners of Brazilian Portuguese (BP) as L3, word order in declarative and in interrogative sentences and the ambiguity of reference in relative clauses. These are domains where English and BP resemble each other more, despite the fact that Spanish and BP are more similar at a general level. The learners were native speakers of English or Spanish, who all had advanced proficiency of either L2 Spanish or of L2 English. Results pointed at transfer from Spanish irrespectively of it being the L1 or the L2, speaking in favour of typology as a strong factor for the selection of background language as transfer source in L3 appropriation.

Syntactic Transfer in L3 Learning. What Do Models and Results …

111

In Rothman (2011) the semantic interpretation of the adjective according to its position in the noun phrase by learners of a Romance L3 (Spanish and Portuguese) was investigated. The typologically closest language (Italian L1 in one group and Spanish L2 in another) was found to be relied upon, this leading to either positive or negative transfer. Another study on Romance languages that supports the idea that closely related languages take on a role as source language of transfer is, for instance, Foote (2009) on tense and aspect. In summary, according to the TPM, neither the L1 nor the L2 can be said to have a privileged status for initial stages of L3 morphosyntactic transfer. Transfer is constrained by the general degree of similarity between languages. It is not entirely clear how the concept of typology relates to that of psychotypology within the TPM. In his first studies on L3, Rothman seemed to acknowledge the learner as being consciously aware of typological relations, as the following quote indicates: Initial State transfer for multilingualism occurs selectively, depending on the comparative perceived typology of the language pairings involved, or psychotypological proximity. Syntactic properties of the closest (psycho)typological language, either the L1 or L2, constitute the initial state hypotheses in multilingualism, whether or not such transfer constitutes the most economical option. To be clear, by ‘psychotypology’ I am referring to a speaker’s perception of typological proximity in the sense of Kellerman (1983). (Rothman, 2011, p. 112)

In Rothman’s subsequent studies, it is posited that initial L3 development is constrained by what the internal parser unconsciously takes to be the most significant structural similarity among the three grammars (L1, L2, and L3), based on an implicationally hierarchical continuum of four types of linguistic cues: Lexicon → Phonological/Phonotactic Cues → Functional Morphology → Syntactic Structure. “Since the TPM predicts that multilingual transfer selection is not dependent on obvious surface level (dis)similarity, it rejects the notion that conscious psychotypological assessment on the part of the learner brings anything to bear” (Rothman, 2015, pp. 184–185).

3.4 The Scalpel Model In an effort to complement previous models on L3 syntax, with one that predicts L3 development, Slabakova (2017) puts forth the scalpel model, which considers the specific properties of linguistic structures as a discriminating factor for transfer: “The model aims to identify and examine what happens beyond the initial state of acquisition and what factors may influence change from one state of knowledge to another” (Slabakova, 2017, p. 651). The author adheres to the idea of integrated multilingualism, that is, that the languages of the multilingual are functionally not separate but integrated (Abutalebi & Green, 2007; Cook, 1991, 1992; Paradis, 2004). This is explained as a “view of the multilingual linguistic competence as an amalgamation of sub-grammars coming from the previously acquired languages, equipped with some sort of differentiation mechanism” (Slabakova, 2017, p. 656).

112

C. Bardel

Slabakova argues against the claim of the TPM of “wholesale transfer” (2017, p. 655), and also against the L2SFH, since, according to the scalpel model, neither the L1 nor the L2 has a privileged status with respect to transfer. As recalled, previous research has shown transfer from either the L1 or the L2 or both. Therefore, the scalpel model predicts potential transfer from all background languages. Slabakova reviews the studies of Bruhn de Garavito and Perpiñán (2014) and Fallah, Jabbari, and Fazilatfar (2016) on L3 beginners and concludes that some structures are learned earlier because of their particular properties. Others require more time, and while they are being appropriated, learners transfer acquired structures from their background languages, either L1 or L2. Bruhn de Garavito and Perpiñán (2014) conducted a study on L3 learners of Spanish who were in their third week of exposure to Spanish. They were L1 speakers of French and proficient L2 speakers of English. Different linguistic properties were tested, for most of which the native language, French, differed from the L2 English and the target L3 Spanish. Results showed that French L1 and English L2 competed as source language of transfer. In some situations the learners relied on their French grammar and in others on their English grammar. These results challenge the idea of whole-sale transfer from one of the background grammars in the first stages of L3 learning. The scalpel model does not offer any answer to the question of what determines which background language will be chosen, and on what grounds. Some studies indicate that learners transfer structures from the typologically closest language, others that they transfer from the language they use more in everyday life. The results from the study by Bruhn de Garavito and Perpiñán (2014) show that single linguistic properties may transfer from the L1, others from the L2, whenever they facilitate the task of processing the L3. The studies reviewed by Slabakova show that transfer from L1/L2 can be detrimental as well as facilitative (Slabakova, 2017, p. 656). Still, it is claimed that “the activated grammatical possibilities of the L1-plus-L2 combined grammar act with a scalpel-like precision, rather than as a blunt object, to extract the enhancing, or facilitative options of L1 or L2 parameter values” (Slabakova, 2017, p. 655). Slabakova argues that there are different learning patterns for different properties, depending on structural conditions. Diverging outcomes on two properties examined in the same learner groups indicate that other language external factors, such as degree of presence in the input (i.e., positive evidence), and degree of correction (i.e., negative evidence) condition learners’ inclinations to transfer. Comprehensible linguistic input is suggested by Slabakova and García Mayo (2017) to be a crucial factor in L3 learning. The bottom line is that L1 or L2 transfer happens property-byproperty and that it is influenced by various factors. To conclude, while reviewing a number of L3 studies with diverging results, with regard to L1 versus L2 transfer, Slabakova highlights that transfer may come from L1 as well as from L2, also in the same learners at a given moment, depending on the structures in question, their frequency in the input, their semantic weight and hence the positive and negative evidence that can be expected. The model highlights several important cognitive and experiential factors affecting transfer that are not taken into account by the other current models (structural complexity, misleading input, ambiguous evidence,

Syntactic Transfer in L3 Learning. What Do Models and Results …

113

frequency, and language activation). This view probably comes close to reality (see De Bot & Jaensch, 2015 for a similar approach), but it is not entirely clear how predictions can be made from this model.

3.5 The Linguistic Proximity Model (LPM) Another recent model on patterns in multilingual syntax is the one proposed by Westergaard et al. (2017). The model, which advocates the role of linguistic similarity for transfer in syntax, builds on data from 22 young 2L1 speakers, that is, 11–14-yearold Norwegian Russian simultaneous bilinguals who had been learning English from the age of six. They were compared to Norwegian-speaking monolinguals (N = 46) and Russian-speaking monolinguals (N = 31), all learning English, in two syntactic tests, one on verb-noun inversion (4) and one on the placement of adverbs (5). (4) What will the little girl read? (5) Susan often eats sweets.

The model differs from the ones previously described here, in that it builds on data from simultaneous child bilingualism. Strictly speaking, the TL is an L2, not an L3, if we define L3 as a language that is learned after the appropriation of another non-native language (see Bardel & Falk, 2012; Falk & Bardel, 2011; Hammarberg, 2010 for discussions on the role of age of onset of the background languages in L3 studies). Similarly to Slabakova (2017), Westergaard et al. (2017) argue for combined (L1/L2) CLI in incremental property-by-property learning and against wholesale transfer from one of the background languages. According to the Linguistic Proximity Model, Ln acquisition allows for both facilitative and non-facilitative influence from one or both previously acquired languages. Cross-linguistic influence occurs when a particular linguistic property in the Ln input reveals abstract structural similarity to linguistic properties of the previously learned languages (Westergaard et al., 2017, p. 670). Norwegian verb placement (verb-second) transfers in both monolingual Norwegian L1 speakers and bilingual Norwegian-Russian L1 speakers into their L2/L3 English, suggesting non-facilitative influence from Norwegian L1 in both monoand bilinguals. Russian L1 seems to have a facilitative effect. The findings indicate that not just typological proximity, but also structural similarity at an abstract level, should be considered an important factor in L3 learning. They indicate that typology plays a role, but not in the sense of TPM, which defines typology as a notion of whole language systems being closely related and similar. Instead, structural typology seems to play a strong role. The young learners of this study transfer (also negatively) from one more closely related language to another (that is, from Norwegian to English) or from one less closely related (from Russian). In the case of Russian L1, knowledge of the particular structure in question is helpful: 2L1 children clearly benefit from Russian, being one of their L1s, which exhibits structural similarity with English in this case.

114

C. Bardel

Turning back to what makes L3 special, the reader will recall the distinction between L1 on the one hand and L2/L3/Ln on the other, made earlier in this chapter in order to claim that the L1, being a native language, differs qualitatively from and L2/L3, the latter being non-native languages. The consequence of this difference in age of onset is that L3 is susceptible to influence not only from the native language, the L1, but also from another non-native language, the L2. For obvious reasons, this view does not apply to cases, in which the L3 label is used to denominate the first non-native language that is learned by bilingual children, with two simultaneously acquired native languages, or L1s (that is, ‘2L1’, as in Westergaard et al., 2017).

4 Discussion of the Five Models During the last decade, a number of empirical studies have appeared aiming at disentangling the different factors claimed to be of relevance for transfer into the L3 (for one most recent, see Puig-Mayenco & Marsden, 2018). Especially, the first three models presented above have been confronted, and some critique has been put forward. Starting with the CEM, it has been supported by follow-up studies on different language combinations, e.g., Berkes and Flynn (2012) on elicited imitation of relative clauses in L2 and L3, and Flynn and Berkes (2017) on L2 and L3 learners’ ability to identify so-called ‘empty categories’ in sentences, that is, nominal elements without phonological content). Although pioneering, the model displays some weaknesses. First of all, it does not account for negative transfer, which is commonly reported as a characteristic of the L3 learning process (see Bardel & Falk, 2007; Rothman & Cabrelli Amaro, 2010, for instance). Secondly, it is not clear what makes the learner select the most appropriate background language as the transfer source for a particular structure, that is, how the learner knows which of the background languages will have a positive effect, without already knowing which is the correct solution in the TL, something which would make transfer unnecessary. Thirdly, the results from the Flynn et al. (2004) study could also be caused by the L2 status factor. As for the L2SFH, some empirical studies on L3 syntax and phonology have lent support to the model (Falk & Bardel, 2011; Forsyth, 2014; Keidel & Hörberg, 2017; Lindqvist & Falk, 2014; Sánchez, 2011a, 2015). However, it has also been contradicted, in particular by studies that advocate the role of typological proximity as a major factor, this typically occurring when L1 and L3 are closely related languages (Rothman, 2010, 2011). The L2SFH in its original version has been revised and modified in Bardel and Sánchez (2017). This modified version of the model accounts for individual variation depending on neurolinguistic and aptitude-related factors, which might mediate the access to metalinguistic awareness and the noticing capacity of the individual learner. As discussed in Bardel and Falk (2012), Falk et al. (2015), and further in Bardel and Sánchez (2017), psychotypology is viewed upon as an inherent trait of explicit metalinguistic knowledge. As mentioned, the TPM has been supported by several studies on L3 learning involving two Romance languages and English (Foote, 2009; Giancaspro, Halloran,

Syntactic Transfer in L3 Learning. What Do Models and Results …

115

& Iverson, 2015a; Rothman, 2010, 2011, 2015), but recently also other language combinations, such as Spanish, Catalan and English (Puig-Mayenco & Marsden, 2019). A legitimate question is whether it is fruitful to compare the models presented above, in order to dismiss one or more of them, presenting new empirical data gathered from learner groups with different language combinations. Hammarberg (2016) points at some complications, especially when comparing the two factors L2 status and typology, in trying to rule any of them out, which is what has been done in several studies during the last decade. For example, Hammarberg remarks that while structural similarity between languages can exist to a greater or a lesser extent, the status of L1 or L2 is a categorical, either-or distinction. Hence, it could be expected that the typology factor will be more likely to trump the L2 status factor when confronting background languages whose degree of similarity to the L3 differs highly, than when confronting background languages that have a more similar degree of distance to the L3. This affects the balance of power between typology and L2 factor in a given constellation of languages. In fact, a problem with the TPM is that it is not clear how similar two languages must be in order for the internal parser to select either the L1 or the L2 grammar as the most relevant source for multilingual transfer. Somewhat inconclusively, and maybe not surprisingly, it can be said that the results from a large number of L3 studies indicate that if one out of two background languages is much more closely related than the other to the TL, this language, whether L1 or L2, will act as an important transfer source. Similarly to the TPM, the scalpel model acknowledges transfer from all background languages, L1 and L2, recognizing also negative transfer caused by typological relations (Slabakova, 2017, p. 657). Still, as we have seen, the model is based on the idea that the interlanguage grammar is able to extract the facilitative options of L1 or L2 parameter values with a scalpel-like precision (Slabakova, 2017, p. 655). Unfortunately, it is unclear how is it possible for this entity to value which options are facilitative, considering that this implies an understanding of the target structure, and consequently the problem with the selectivity of the CEM is not solved. That transfer can be detrimental has been shown in, for example, Bardel and Falk (2007), Falk and Bardel (2010), and Rothman and Cabrelli Amaro (2010), as well as subsequent studies by Rothman, as quoted above. Further evidence on detrimental transfer from the study by Slabakova and García Mayo (2015) is quoted by Slabakova (2017). However, Slabakova (2017, p. 659) downplays the role of typology by claiming that “crosslinguistic influence can be due to linguistic, experiential, and input factors, not only typology”. In fact, the diverging results from different studies could also be accounted for by other factors than the L2 status factor and the typology factor, such as proficiency level in the L2 as well as in the L3, aptitude, modality of use, and affective factors (motivation, language prestige, etc.). For example, proficiency in the L2 was shown to be an important factor in L3 learning of syntax in a recent study by Sánchez and Bardel (2017), who examined negative transfer in syntax from German L2 into English L3 in young bilingual L1 speakers of Catalan and Spanish. The learners had different levels of proficiency in English L2 and the results showed that learners with low and intermediate L2 proficiency

116

C. Bardel

displayed more negative transfer than those learners who had an advanced level of proficiency in their L2. As seen, the reviewed models predict differently as regards whether L1 or L2 will have a privileged position with respect to influencing the L3 grammar. While the L2 status factor, based on the high degree of metalinguistic knowledge, is given a strong role in the L2SFH model, and the typologically closest background language in the TPM, no previously learned grammar has an a priori privileged role in the CEM, the LPM or the scalpel model. Instead, predominance of communicative use, hence activation, may certainly modulate the influence from previously acquired languages, as suggested by Slabakova (2017, p. 656 and Sect. 3.4 in this chapter). This idea is backed up with the study by Fallah et al. (2016) referred to by Slabakova, which showed that the language that was used more in daily life was the one that influenced the L3 most, indicating that language dominance may be a discriminating factor in cases of high level proficiency of the L2. For another study that found similar results, see Neuser (2016), who found, in the case of lexical transfer, that the language used most in oral communication in daily life is a significant predictor for source language of transfer in spoken L3 production, while the language used most in writing in daily life was a significant predictor for the source of transfer in written L3 production. Considering the tenets underlying some of the hypotheses on non-native language learning presented above, it is interesting from a theoretical point of view to try to capture the very first hours of L3 language learning. Still, very few studies look directly into the true L3 initial state (Flynn & Berkes, 2017, pp. 35–36; Schwartz & Sprouse, 1996). As pointed out by Westergaard et al. (2017, p. 669), referring to Rothman’s TPM, “since the choice of the transfer source depends on the comparison between certain structural properties of the L3 and the two previously learned languages, transfer cannot occur at the initial state, before any exposure to the L3. Instead, the learner needs to have some basic experience with the L3 before the parser can make the decision”. This implies that a certain level of development is required for the learner to be able to notice, analyze and recognize traits in the input that correspond in previously acquired languages. A study of L3 syntax that comes close to saying something about both the L3 initial state and subsequent development is the one by Bardel and Falk (2007), where dominance of L2 transfer over L1 transfer was found at the very first encounter with the L3. The study followed a language class longitudinally from the very first lesson, and showed that after a period of massive transfer from L2 to L3, the structure in question—sentence negation—was successively learned. Furthermore, Falk and Bardel (2011) examined the placement of object pronouns in L3 at intermediate levels of proficiency and found evidence for L2 transfer there as well, though less distinctively than in the initial state. It is plausible that both L1 and L2 transfer will have to give way for learning the target structures when learners start noticing and internalizing them. While doing so, learners will either automatize learned knowledge or acquire competence implicitly, depending on various individual affective and sociolinguistic factors, such as motivation, degree of exposure, etc. (Paradis, 2009). Studies of the L3 at more advanced levels of development are, for instance, represented in the body of research on English L3 in the Basque Autonomous Community

Syntactic Transfer in L3 Learning. What Do Models and Results …

117

that are reviewed in García Mayo and Rothman (2012). Furthermore, also the study by Slabakova and García Mayo (2017) on topicalization regards rather advanced English L3 speakers in that context.

5 Implications for Language Learning and Teaching What conclusions can teachers draw from research into L3 on transfer from previously acquired/learned languages, in terms of pedagogical implications? Before trying to answer this question, a few words of caution may be in order. It is not obvious that SLA research, nor research specifically devoted to third language acquisition/learning, is always applicable to language pedagogy. Although giving some examples of possible implications that we will return to below, Ringbom and Jarvis (2009, p. 114) observe that there is still a long way to go before researchers will have enough knowledge about language learning mechanisms to be able to pronounce themselves in a definitive way about how languages are most effectively taught. It has also been pointed out by Spada (2015, 2018), that SLA research is not always, and cannot always be, applied in a way that is relevant for language teachers. Many of the questions raised by researchers are different from those asked by teachers and the ways in which researchers seek answers to their questions often differ from classroom practice (Spada, 2018). It is also important to acknowledge that language teachers have their own experience-based cognitions about what to teach and how to teach and perceptions of how learners learn what is being taught (Borg, 2015). Therefore, there often remains a gap between the results of SLA studies and the suggestions that researchers may have for language teachers on the one hand and the real needs present in the language classroom on the other. It is, however, one of the main tasks of applied linguists to communicate their results to the intended audience; for a successful example, see Little (2016). There are a few attempts to convert results from L3 research as well as general SLA research on syntax into teaching methods (see part III in Angelovska & Hahn, 2017), and for a recent special issue on the acquisition of second language grammar and language pedagogy, see Marsden and Slabakova (2018). However, it has turned out to be relevant and useful for both teachers and researchers to conduct research into the language classroom in collaboration, that is, research projects with teacher participation or researching teachers with supervision of researchers (for examples, see Bardel, Erickson, Granfeldt, & Rosén, 2017). In an attempt to provide some suggestions for areas of investigation, we will turn back to the five models of L3 syntax that have been examined in this chapter and discuss their relevance and implications for language education. Starting with the CEM, we recall that this model claims that all previously acquired or learned languages (L1, L2) play a positive or neutral role in further (L3) learning. The empirical results that the model is based on indicate that learners may draw advantages from their multilingual repertoire, especially when it comes to syntactic structures. This is, of course, an important point of departure for teachers in the multilingual classroom. Tavakol

118

C. Bardel

and Jabbari (2014) argue that there is no point in trying to repress the other languages known to the learner, given that all languages are integrated and interact. Rather, it may be a good idea to work explicitly with comparisons between languages, drawing on metalinguistic knowledge in order to raise metalinguistic awareness. Empirical classroom studies in this area would help shed more light on this assumption. However, later studies on L3 syntax indicate that all previous language knowledge is not helpful tout court. When it comes to specific structures that differ between L2 and L3 (cf. the L2 status factor) or between closely related and generally similar languages (cf. the TPM), transfer may lead to the wrong option when the learner makes hypotheses about the L3 or expresses herself or himself in the L3. However, it has not yet been shown that this negative transfer lasts for longer periods, nor that it has a generally detrimental effect on the communicative competence of the L3 speaker. Therefore, in this case, the task of the teacher may be to make the learner notice the small error (for instance, in word order) and, if possible, point out similarities and differences between the languages in question. In their chapter on the role of cross-linguistic similarity in second language learning, Ringbom and Jarvis (2009, pp. 114–115) argue that learners look for all kinds of similarities wherever they can find them. They also note that when learning a closely related language learners have a relatively small learning burden because they will easily find similarities. If a background language and the TL are closely related, an ideal task of the teacher would be to systematically outline recurring similarities, but the closer the relation between the languages, the more important it becomes also to point out unexpected differences. This, however, puts a lot of responsibility on the teacher, who cannot be expected to know all languages of his or her students’ repertoires. The more recent models, the scalpel model and the LPM, point to a number of other factors (in addition to typology) that may be of relevance for negative and positive transfer from the background languages, and therefore may play a role in L3 development. It is premature, however, to draw pedagogical conclusions from these two models. More studies are needed. Both metalinguistic knowledge and learners’ capacity of noticing similarities and differences between languages are crucial notions in language education. The teacher plays an important role for efficient language learning through consciousness-raising and noticing linguistic form (Oliveira & Schleppegrell, 2015). The L2SFH builds its tenets on the notion of metalinguistic knowledge. Following the D/P model, it is acknowledged that humans have two different memories that take care of language: declarative and procedural memory. Everyone has access to both types, but procedural memory plays a crucial role in L1 acquisition of syntax, tends to be dominant in L1 syntax, and becomes slightly weakened after childhood, while declarative memory gets stronger during childhood and is at its best in early adulthood (Ullman & Lovelett, 2018). When learning a second or foreign language later in life, declarative memory plays a more important role, especially in formal learning contexts, subserving explicit metalinguistic knowledge. A basic claim of the L2SFH is that L1 acquisition and L2/L3 learning differ fundamentally from a qualitative point of view, in the sense that while much of our L1 verbal communicative capacity relies on implicit competence, L2/L3 mastery is built upon metalinguistic knowledge (Par-

Syntactic Transfer in L3 Learning. What Do Models and Results …

119

adis, 2009). This has implications for language transfer. However, metalinguistic knowledge can occur also in the L1 (knowledge about words and their meanings, knowledge about grammar or rules of pronunciation, etc.) and therefore in additional language learning, all previous language learning matters. Hence, we rely on declarative memory also when using our L1 in specific functions. As shown in Falk et al. (2015), a high degree of explicit metalinguistic knowledge about the L1 grammar can lead to positive transfer of structures from the L1 into a L3. This is important for the teacher to take into account when working with different individuals. Variation in success among L3 learners is a phenomenon well-known to teachers that can be explained within the D/P framework. Metalinguistic knowledge and awareness are crucial components for L3 learning, and learners manage to different extents to rely on and take advantage of this. This is related to the fact that there is individual variation regarding the ability to access important components of declarative memory such as working memory, attention, control, etc. (Bardel & Sánchez, 2017; Paradis, 2009). In relation to this, a question that becomes relevant for language teachers is whether metalinguistic knowledge can be enhanced in the classroom. Given the D/P framework, this must be one of the most important tasks of the teacher, together with offering suitable input. The input and the learners’ chances to process it is probably influenced by factors such as noticing capacity and awareness of language relations (Sánchez & Bardel, 2017), and it is an empirical question whether these factors can be enhanced in the classroom. Moreover, Ullman and Lovelett (2018) argue that at least two typical and feasible classroom activities can enhance learning, benefitting mainly declarative, but also procedural, memory: spaced repetition (also known as distributed practice), which means introducing temporal gaps between repeated presentations of the same item, and retrieval practice (also known as the testing effect), meaning retrieving learned information from memory instead of restudying it. This kind of memory enhancing work is an important task for a language teacher, together with offering suitable and processable input.

6 Conclusion This chapter aimed to discuss some attempts to answer the question of how previously acquired or learned languages—first, second or foreign languages (L1/L2/FL)—play a role in the learning of a L3/Ln, especially syntax. A secondary aim of the chapter was to examine what research into L3 syntax can offer language teachers. Five existing models of L3 syntax have been surveyed and their relevance and implications for language education have been discussed. It is clear from the complete body of research into L3 syntax that in new language learning, all previously learned knowledge about languages and all acquired language competence matter. This is important for the teacher to take into account when working with different individuals. Explicit metalinguistic knowledge is a crucial component for the development of a non-native language, be it L2, L3 or Ln, and such

120

C. Bardel

knowledge about previously learned second or foreign languages can be exploited in an informed way, by teachers, as well as learners. Previously acquired implicit linguistic competence is a resource that can constitute a basis for developing explicit knowledge in parallel and becoming aware at a metalinguistic level about how one’s mother tongue works. In order to make this resource interesting for the language classroom, learners can and will, in a best-case scenario with help from their teachers, notice and become aware of the L1 grammar and phonology, in order to be able to develop useful knowledge for additional language learning in the classroom. No teacher can be expected to know all the background languages present in a multilingual classroom. But all teachers can be aware of the number of background languages and students’ level of proficiency in them as well as of the individual variation among their students, and put in support where needed. All teachers can help to raise students’ metalinguistic awareness about typological relations among their L1, L2, and L3 in order to enhance positive transfer and avoid negative transfer. All teachers can help encourage learners to notice similarities and differences, and work with this continuously at all levels of development of the third language.

What Lies Beneath: L1 Morphosyntax Seeping in Through Young Learners’ EFL Raúl Azpilicueta-Martínez

Abstract Recent findings regarding L1 use among students learning English as a Foreign Language (EFL) have called into question the notion of a negative correlation between proficiency and lexical crosslinguistic influence (CLI). However, interaction-based studies often focus on CLI via explicit L1 use exclusively. While this study also analyses explicit L1 use, it primarily taps into the underlying impact of L1 morphosyntax shaping part of the students’ oral production in L2, a fact which has already been pointed out in the literature, yet needs looking into in further detail. Participants were 20 children aged eight, L1 Spanish beginner learners of English in a Content and Language Integrated Learning (CLIL) programme. They took on the narrator role in a communicative task which they carried out firstly with (i) an expert speaker of English and, subsequently with (ii) a matched-level peer. The results boast an extremely low rate of explicit L1 use, yet reveal the existence of more pervasive CLI in the form of structural transfer, particularly so when children interacted with their peers. This fact might hint at a trade-off strategy for their low mastery of the target language. Keywords Children · CLI · CLIL · EFL · Interaction · L1 use · Task

1 Introduction Most research on oral interaction in Second Language Acquisition (SLA) spawned from the interactionist framework (Long, 1996), and has defined the features and learning opportunities of learners’ negotiations considering a myriad of variables, such as age, target language (TL) level, context and task-type, to name but some. Research analysing the benefits of interaction for SLA initially focussed on adult learners (Gass & Varonis, 1985; Pica & Doughty, 1985; Porter, 1986; Yule & R. Azpilicueta-Martínez (B) Departamento de Ciencias Humanas y de la Educación, Universidad Pública de Navarra (UPNA), Edificio Los Magnolios-Campus de Arrosadía, 31006 Pamplona, Spain e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_7

121

122

R. Azpilicueta-Martínez

Macdonald, 1990). Subsequent studies by Alison Mackey and Rhonda Oliver (Mackey & Oliver, 2002; Mackey, Oliver, & Leeman, 2003; Oliver, 1995a, 1995b, 1998, 2000, 2002; Oliver & Mackey, 2003; Philp, Oliver, & Mackey, 2008) included children learning English as a Second Language (ESL). However, data from English as a Foreign Language (EFL) children remained non-existent until more recent research (Azkarai & Imaz Agirre, 2016; García Mayo & Lázaro-Ibarrola, 2015; Lázaro-Ibarrola & Azpilicueta-Martínez, 2015; Lázaro-Ibarrola & Hidalgo, 2017; Philp & Tognini, 2009; Tognini, 2008; Tognini & Oliver, 2012). Regarding EFL and Content and Language Integrated Learning (CLIL) interaction-based research, some of the commonly addressed features are negotiation for meaning strategies (including conversational adjustments and repetitions) and explicit L1 use (e.g., L1 borrowed words). The latter element is undergoing a shift in the way it is perceived, given the studies showing the benefits of a balanced explicit L1 use in interaction (García Mayo & Hidalgo, 2017; Lázaro-Ibarrola & García Mayo, 2012; Macaro, 2005; Neokleous, 2017). A recent study by LázaroIbarrola and Hidalgo (2017) warned about the occurrence of abundant L1 structural transfer (e.g., L1 literal translations, L1 plural forms) in their data (young L1 Spanish EFL learners engaged in oral interaction, a similar setting to the present study) and advocated the inclusion of this phenomenon as an object of study in subsequent interaction-based studies (e.g., García Mayo & Hidalgo, 2017; Lázaro-Ibarrola & Hidalgo, 2017). This study attempts to shed light on the oral CLI of 20 young L1 Spanish learners of English enrolled in a CLIL programme while performing a picture-based task which embeds a spot-the-difference1 activity within a story. Specifically, we will analyse all instances of explicit L1 use and structural transfer in their interactions.

2 Literature Review 2.1 L1 Use: Proficiency and Age Young learners (hereinafter ‘YLs’) have been reported to need to resort to the L1 while producing in the TL at early stages of its acquisition (Agustín-Llach, 2009; Celaya & Torras, 2001; Gabrys-Barker, 2006; Gost & Celaya, 2005). This has been attributed to lexical lacunae in their L2 coupled with the need to communicate in the TL, and is widely regarded as a common compensatory communication strategy (Celaya, 1992; Ecke, 2001; James, 2013). However, explicit L1 use among students still appears to be frowned upon by certain educators on the grounds that learners will fail to perform the tasks at hand in the TL (Storch & Aldosari, 2010; Tognini & Oliver, 2012), in spite of the substantiated benefits of a balanced explicit L1 use reported in the literature (Antón & DiCamilla, 1 See

Sect. 4.2.

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

123

1998; DiCamilla & Antón, 2012; García Mayo & Hidalgo, 2017; Lázaro-Ibarrola & García Mayo, 2012; Macaro, 2005; Neokleous, 2017; Storch & Aldosari, 2010). Different scholars have suggested that proficiency in the TL and explicit L1 use show a negative correlation (Agustín-Llach, 2009; Herwig, 2001; Jarvis & Pavlenko, 2008; Navés, Miralpeix, & Celaya, 2005; Odlin & Jarvis, 2004; Serra, 2007; Storch & Aldosari, 2010). In other words, explicit L1 use appears to be inversely proportional to the mastery of the TL. However, a number of studies have cast doubt on such clear-cut interrelationship (e.g., Sánchez, 2003; Sanz, 2000). Recently, a growing body of interaction-based research has compared, among other variables, explicit L1 use in EFL and CLIL YLs (Azkarai & García Mayo, 2017; Azkarai & Imaz Agirre, 2016; García Mayo & Hidalgo, 2017; García Mayo & Lázaro-Ibarrola, 2015), with mixed results hinting at additional factors such as motivation and attitude (García Mayo & Lázaro-Ibarrola, 2015, p. 49) playing a central role in the students’ degree of reliance on their L1. On the one hand, recent studies have reported higher levels of L1 terms in mainstream (hereafter MS) EFL learners than their age-matched, more proficient CLIL counterparts (Azkarai & Imaz Agirre, 2016; García Mayo & Imaz Agirre, 2017; García Mayo & Lázaro-Ibarrola, 2015), supporting previous claims by Möhle (1989), Poulisse (1990), and Poulisse and Bongaerts (1994). By contrast, the latest research by Pladevall-Ballester and Vraciu (2017) revealed similar rates of explicit L1 use in both instructional settings (CLIL and MS EFL), while a study by Lázaro-Ibarrola and Azpilicueta-Martínez (2015) reported a minimal percentage of explicit L1 use (0.52%) in MS EFL learners with a considerably low command of the TL. Interestingly, recent research has also revealed opposing results regarding age and explicit L1 use, with studies reporting greater explicit L1 use by the older students (e.g., García Mayo & Hidalgo, 2017; García Mayo & Lázaro-Ibarrola, 2015), while Azkarai and Imaz-Agirre (2016) (MS EFL) and Pladevall-Ballester and Vraciu (2017) hint at the opposite, concurring with previous work by Pinter (2006). García Mayo and Hidalgo (2017) point at “learner motivation, task complexity, task repetition and instructional setting” (García Mayo & Hidalgo, 2017, p. 133) as key elements impinging on the amount of L1 in the learners’ output. Likewise, and, in spite of the low L1 rates in some of the research above, the same authors (García Mayo & Hidalgo, 2017; Lázaro-Ibarrola & Hidalgo, 2017) have brought into light the existence of abundant L1 ‘structural transfer’ (García Mayo & Hidalgo, 2017, p. 141) in their data, understood as ungrammatical instances of the TL. Non-explicit CLI with Spanish learners of English has already been object of study. For example, Celaya (1992) analysed the way the L1 (Spanish and/or Catalan) affected the TL of secondary-school students via structural L2 choice (Celaya, 1992). Among other things, she found out how the Spanish and Catalan tense systems may lead to an overgeneralisation in the use of the simple present in English when referring to present time, as a consequence of the “different distributions of forms and meanings in English and Spanish and Catalan tense-systems” (Celaya, 1992, p. 29). The present work, specifically addresses the way the L1 might lead the student’s production to produce ungrammatical forms in the TL, which would otherwise constitute grammatically valid utterances in the L1 if translated word for word.

124

R. Azpilicueta-Martínez

2.2 L1 Use: The Interlocutor and Task Factors The interlocutor factor is one of the aspects subject to affect the amount and type of interaction in a SL or FL at different levels. Circumstances such as age, gender, personality or status have long been investigated in the literature (Berry, 1997; Buckingham, 1997; Porter, 1991; Porter & Shen, 1991; to name but some), and are ultimately thought to exert some influence as regards language choice (e.g., Allwright, 1996; Bell, 1984). Regarding low-level children, research by Lyster and Izquierdo call into question the suitability of content-based tasks with young children whose command of the TL might be too low (Lyster, 2001; Lyster & Izquierdo, 2009). An empirical study by Lyster (2001) with young students (ages eight-ten) learning French in an immersion context, claims that, in communicatively oriented classrooms “young L2 learners may not readily notice target-non target mismatches in the interactional input” (Lyster, 2001, p. 268). Lyster’s study, therefore, seems to indicate that children are perhaps able to interact but may not do so in a way that results in accuracy. A study by Broner (2001) investigated, among other elements, the interlocutor and task factors in relation to the amount of TL use with ten-year-old students in a Spanish immersion context in the US. She concluded that interlocutor, task-content and “being on/off task” all played central roles in the amount of TL and L1 produced by the students. By being ‘on-task’ Broner referred to language use “in which the children are speaking about and are engaged in the assigned task”, whereas ‘off-task’ language included language tokens “in which the children are not carrying out the assigned task” (Broner, 2001, p. 126) She pointed at the adult expert speaker of TL Spanish holding the teacher role in her study as directly boosting the students’ output in that language, yet also explained how activities which forced the students to focus on the TL in peer-peer interaction, such as creative writing, led to a proportionally higher TL output (though not entirely L1-free) than other types of activities in non-languagerelated contents (e.g., maths and science). In addition, Broner’s study revealed a significantly higher number of L1 items during off-task communication, as opposed to García Mayo and Hidalgo (2017), whose results displayed no instances of explicit L1 use in off-task interaction whatsoever. A recent study on peer-peer interaction by Lázaro-Ibarrola and Hidalgo (2017) delved into the benefits and limitations of conversational interactions among 11-year old learners of English in a CLIL context which was very similar to the one in the present study. While not central to their study, these authors report abundant instances of CLI via structural transfer, and explain that (i) such instances never triggered recasts or explicit corrections from their interlocutor, i.e., they were likely to be understood by their counterparts, and (ii) such structural transfer appeared to be ‘mirrored’ in paired interaction, that is, when one of the interlocutors started to use a given L1 pattern in the TL, such as the omission of the auxiliary in questions, then that pattern became recurrent along their exchanges, as the following example (1) illustrates: (1) 1 *CHI1: You have a girl that his hair is yellow? 2 *CHI2: No, you?

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

125

3 *CHI1: No. 4 *CHI2: You have a little girl that have flowers in his t-shirt in the in the (…)? 5 *CHI1: Playground? 6 *CHI2: Yes. 7 *CHI1: Yes I have it. 8 *CHI2: Now you. 9 *CHI1: You have a boy that his t-shirt is red white and blue? (Lázaro-Ibarrola & Hidalgo, 2017, p. 18).

Regarding task choice, several scholars investigating interaction using picturebased prompts have highlighted the importance of choosing age and interestappropriate tasks, since a lack of motivation is hypothesized to lead to a higher explicit L1 use in more proficient FL learners (García Mayo & Hidalgo, 2017; Lázaro-Ibarrola & Hidalgo, 2017), as mentioned above. One type of referential task commonly used in the elicitation and assessment of TL samples are picture stories (e.g., Cambridge Young Learners examinations). Duff, Rossiter, Derwing, and Jones (2008) proposed a series of 33 evaluation criteria for the selection and creation of effective picture stories for L2 research, and they focussed on the nature and structure of the stories, the cultural content in them, as well as on the technical specifications of the pictures themselves. One of the aims of these authors was to outline a useful framework in order to free stories from “confusing visual elements so that participants can focus on the linguistic demands of the story” (Duff et al., 2008, p. 326). However, these authors acknowledge the fact that the efficacy of a given story might depend on the linguistic phenomenon under study, i.e., a given task might be useful in eliciting verb tenses yet fail to provide phonological evidence of certain phonemes when the key words including those phonemes are unknown by participants, and, consequently, avoided in their production. In relation to the application of picture stories in CLI research, Sánchez and Jarvis (2008) concur and note how the suitability and effectiveness of a given visual prompt ought to be assessed according to the CLI phenomenon under study. Furthermore, these authors explain how picture stories which might be deemed as unclear or troublesome might help surface different mental conceptualizations and categorizations in learners from different native-language backgrounds (Sánchez & Jarvis, 2008, p. 330) which would otherwise remain not visible. The present study will also reflect on the possible impact of the violation of some of the criteria in Duff et al. (2008) on the output of students sharing their mother tongue. Finally, according to the ideas mentioned above, the suitability of a given visual task should also be explored in light of the amount and type of TL and CLI generated, an aspect this work attempts to provide insights on (see Sect. 5.2.1).

126

R. Azpilicueta-Martínez

3 Research Questions In light of previous findings, the following research questions were posited: 1. How does the L1 use in young CLIL learners of English compare to results in previous interaction-based studies? 2. What forms of L1 Spanish structures are present in the English interlanguage of young CLIL learners of English when performing interactive oral tasks? 3. To what extent do the interlocutor and task factors impinge on the amount and type of CLI in young CLIL learners of English sharing Spanish as the L1?

4 The Study 4.1 Participants Twenty (20) children—11 girls and nine boys—participated in the present study. They were studying their Year Three course (eight-nine years old) at a state school located in Pamplona, a city in Northern Spain. They all shared Spanish (or Spanish plus another language different to English) as their L1 and had limited access to English-speaking interaction outside their classes. All their teachers had a minimum certified C1 (Common European Framework of Reference for Languages, CEFR) level of English. The proficiency of spoken English for the children in this study was based on the school’s internal assessment records in the EFL subject, as well as on their performance in the diagnostic tests carried out by external examiners from the regional administration the previous year, tests which the author of this paper took part in designing and which had placed the students’ oral proficiency level at pre-A1/A1. The use of additional forms of assessment, e.g., external picture-based oral proficiency A1 test such as Cambridge’s YLs tests, were discarded in order to avoid an impact in terms of familiarity with the task or the interlocutor (see Chambers, Galaczi, & Gilbert, 2012, for further information). Likewise, written placement tests, e.g., Oxford Quick Placement Test, were not considered suitable for the assessment of the oral proficiency of the students, especially given the likely mismatch in proficiency between the students’ oral and written skills (Geva & Shanahan, 2006). Inevitably, this constitutes one limitation to the present study. Neither high nor lowperforming students were included in order to guarantee maximum homogeneity in their proficiency levels. The researcher in this study (labelled as ‘expert’ following the terminology frequently used in SLA research for the sake of clarity) is an L1 Spanish proficient speaker of English, he obtained Cambridge’s Certificate of Proficiency in English (CEFR C2) including the maximum scoring in the oral skill, and has vast experience

What Lies Beneath: L1 Morphosyntax Seeping in Through Young … Table 1 Participants’ profile

127

CLIL children Average age (mean)

8.5

English proficiency

Pre A1/A1

School year

Year three primary education

TL hours of exposure/Week

Ten

both as an EFL teacher and oral examiner with children and adults.A summary of the participants’ characteristics related to age and proficiency can be seen in Table 1.

4.2 The Task The tasks in the present study were designed to elicit the production of oral output. They embedded a spot-the-difference referential task within a storytelling activity, and included the following layout: one of the participants (‘narrator’ role) was provided with a story which had been arranged sequentially in five pictures (see Appendix, ‘Story as seen by narrator’), while their counterpart (‘story builder’) was given eight jumbled-up pictures which included the ones in the story plus three distractor pictures (see Appendix, ‘Story builder’s sheet’), which were similar but not identical to those in the narrator’s story. One subject narrated the story to the other so that the latter had to arrange the story chronologically and leave the three wrong pictures out. An opaque screen was placed between students in order to minimize non-verbal communication. A detailed description of the task can be seen in Table 2.

4.3 Procedure Data collection took place between 14th and 23rd April 2015. All parents were informed that their children’s performances would remain anonymous and limited for research purposes exclusively. Due permission was granted by both parents and the school itself. At the beginning of the study children were told that they were going to take part in a game in English. They were reassured that this was not a test or examination of any sort. Children pairing was semi-random following alphabetical order. Before pairing the students, teachers were asked whether they reckoned any students ought (or not) to be paired in any particular way but no students were deemed particular consideration in this respect. It is noteworthy to point out that students did not know that the researcher could speak their mother tongue. While he did not say so to the children explicitly, he always addressed them in the TL so as to maximise their output in English, and students might have taken it for granted that he was a

128

R. Azpilicueta-Martínez

Table 2 Task description Story-based picture placement task Student

Narrator

Story builder

Description

Without showing their partner their pictures, they must narrate/describe a story in order for their peers to place the story in the right order and leave the wrong distractor pictures aside

Without showing their partner their pictures, they must interact with the narrator so that they can place the story in the right order and leave the wrong distractor pictures aside

Type

Mixed: Information Gap Task + Storytelling

Flow of information

Two-way

Exchange of information

Required

Outcome

Closed (Learners had to end up arranging the story in a specific order)

Use of language

Open (The language used is unpredictable and might vary significantly)

monolingual speaker of English. The data from each participant were collected at four times as follows: Day 1: Tasks 1 and 2 (learner-expert) At time 1 (henceforth Day 1), every participant performed two of the story-telling tasks with the expert (the researcher). Firstly, the expert narrated the story to the learner, who had to build up the story (D1T1). This arrangement intended to create participant-friendliness so that the students could get acquainted with the task with a minimal amount of explicit instructions required. An interlocutor frame was used to guarantee the same degree of input in all cases. Immediately after finishing D1T1, the expert and the student swapped roles and the latter acted as narrator, while the expert had to build up the story. Day 1. Task 1 (D1T1). Expert narrates to learner. Day 1. Task 2 (D1T2). Learner narrates to expert. Day 2: Task 1 and 2 (learner-learner) Seven (7) days later, at time 2 (hereinafter Day 2) the tasks exclusively involved student-student interaction. Firstly, one of the students narrated the story to their story-building partner (D2T1), while they swapped roles afterwards (D2T2). Day 2. Task 1 (D2T1). Learner B narrates to learner A. Day 2. Task 2 (D2T2). Learner A narrates to learner B. These tasks outlined two different participant roles—‘narrators’ and ‘story builders’—depending on the task at hand, as noted Table 3.

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

129

Table 3 Tasks and role distribution Story-based picture placement task Day 1 Task 1 (D1T1) Birthday Party

Researcher as narrator

Seven-day interval

Day 2 Task 1 (D2T1) On a Rainy Day

All students as story builders Task 2 (D1T2) The Snowman

All students as narrators Researcher as story builder

Student B as narrator Student A as story builder

Task 2 (D2T2) The Toyshop

Student A as narrator Student B as story builder

4.4 Data Coding All of the participants’ interactions were coded independently by the author of the paper and an additional researcher, and subsequently compared. Inter-rater reliability was calculated using simple percentage agreement, which resulted in 92%. Any remaining discrepancies were solved individually on a case-by-case basis.

4.4.1

L1 Use

Different scholars have established different categorisations in order to code lexical CLI in their research (e.g., Alegría de la Colina & García Mayo, 2009; Hammarberg, 2001; Ringbom, 2001). After identifying the instances of such items in our transcriptions, we found out the students’ explicit L1 use fit within the following two categories: ‘Edit’ and ‘Insert: explicit elicit’ (Hammarberg, 2001). In the present study, the former covers forms of self-repair (similarly coded as ‘discourse markers’ in Alegría de la Colina & García Mayo, 2009), while the latter refers to explicit switches in order to elicit the English expression from their interlocutor (coded as ‘Vocabulary: deliberation over vocabulary’ in Alegría de la Colina & García Mayo, 2009). These functions are illustrated in the following examples from our participants: EDIT (2) 1 *CHI1: The boy put his…erm… Ay! (English: Oh!) His… Jacket and the girl is putting, erm… too the raining boots, pink. (…) I do next? [EDIT] INSERT: EXPLICIT ELICIT (3) 1 *CHI1: And, how do you say ‘van anciano’ (English: ‘go grandfather’)? [INSERT: EXPLICIT ELICIT].

130

R. Azpilicueta-Martínez

4.4.2

L1 Structures

The present study uses the term ‘L1 structures to refer to instances of L1 Spanish morphosyntax emerging from the data, i.e., underlying Spanish morphology and syntax features seeping through the participants’ output in the form of ungrammatical English, as well as lexical CLI other than explicit L1 use. With the aim of adding insights into the research questions, all transcriptions were coded using a scheme that emerged from the data, which led to the categorisation below, illustrated by examples from the present study. Language tokens which have already been categorised in the literature as developmental errors in the L1 acquisition of English have not been included in our results, such as the following examples found in the data: (4) 1 *CHI1: The girl put the eyes of the snowman.2 (Dulay & Burt, 1974a, p. 134) (5) 1 *CHI1: And he eat a banana.3 (Dulay & Burt, 1974a, p. 132)

Likewise, ungrammatical instances which would entail an ungrammatical form if translated into Spanish have also been discarded: (6) 1 *CHI1: Girl and boy are imagine in the toyshop. (Spanish: ‘Niña y niño están imagina en la tienda de juguetes’.) (7) 1 *CHI1: Boy is take off the… jacket. (Spanish: ‘Niño está quita la… chaqueta’.) (8) 1 *CHI1: You doesn’t say the supermarket. (Spanish: ‘Tú no dice el supermercado’.)

Thus, only those instances of ungrammatical English whose word-for-word translation into Spanish would provide a grammatically correct language token have been included. Translations for each example have consequently been provided. The bespoke classification below encompasses all such instances, some of which fully match well-documented CLI phenomena common in the acquisition of L2 English by non-native learners (Examples 12, 13). However, the data analysis also revealed language tokens which blended features of already reported CLI phenomena with morphosyntactic characteristics unique to the Spanish language (Example 6). In all cases, a reference will be provided. Elision. This category comprises the omission of single terms such as dummy subjects, prepositions, subjects, as well as the elision of multiple words within a sentence (coded as ‘other forms: expanded): (9) 1 *CHI1: In one picture are balloons black and the other red [Elision: Dummy subject] (Spanish: ‘En un dibujo hay globos negros y los otros rojos’.)

for Structure 6: (NP-Aux) − V + ing − (Infin)-Nop-Prep-NP/NP-NP. count for structure 1: NP − V-Pron.

2 Count 3 Error

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

131

(10) 1 *CHI1: Putting the hat [Elision: preposition] (Spanish: ‘Poniéndo (se) el sombrero’.) (11)4 1 *CHI1: Is jumping in the bed? [Elision: subject] (Spanish: ‘¿Está saltando en la cama?’) (12) 1 *CHI1: About that they don’t have (…) [Elision; other forms: expanded] (Spanish: ‘Sobre que no tienen (…)’.)

L1 plural forms. Here we include the pluralisation of terms which only exist in its singular form in English, and they were found in irregular plural words in English (i.e., not ending in ‘s’) and singular numeral determiners: (13) 1 *CHI1: Childrens [L1 plural forms: double plural] (Spanish: ‘Niños(s)’.) (14) 1 *CHI1: ‘Ones balloons’ [L1 plural forms: Singular numeral determiner] (Spanish: ‘Uno(s) globos’.)

Overgeneralisations. This concept refers to the ungrammatical use of terms or structures whose meaning or usage is more limited or less general than stated. In the present study it includes gender nouns, particles ‘no’ or ‘yes’ as substitutes for contracted forms and possessives: (15) 1 *CHI1: The two boys (girl and boy) are… are… bored [Overgeneralisation: gender] (Spanish: ‘Los dos niños (masculine) están… están… aburridos’.) (16) 1 *CHI1: Here are two balloons and here no [Overgeneralisation: negation particle] (Spanish: ‘Hay dos globos y aquí no’.) (17) 1 *CHI1: Two children and his (their) father 5 [Overgeneralisation: possessive] (Spanish: ‘Dos niños y su (‘su’: ‘his/her’ but also ‘their’) padre’.)

Additional types of transfer. Here are included further instances of CLI spotted in our data, including subject-adjective inversion, transfer of L1 prepositions and syntax at phrase level, subject-verb inversion in questions, circumlocution and calques6 : (18) 1 *CHI1: Balloons black and the other red 7 [Additional: subject-adjective inversion] (Spanish: ‘Globos negros y los otros rojos’.) 4 Example (11) features declarative word order, no inversion and no fronting, i.e., a Stage 2 question

structure according to the L2 acquisition of English questions in Pienemann, Johnston, and Brindley (1988), as opposed to the French L1 subjects in Pienemann et al’s (which would correspond to ‘He is jumping in the bed?’). By contrast, Example (11) also includes subject omission, which could constitute a variation of Stage 2 questions specific to Spanish L1 learners. 5 Possessive Determiners, Stage 2: Emergence on French-speaking students (White, 1998). 6 In the present study ‘circumlocution’ refers to roundabout expressions in the TL which, translated word-for-word would lead to a similar expression in the L1, i.e., it is not possible to determine whether the learner knows the specific term in the L1. ‘Calques’ in this study includes borrowed, literal word-for-word translations. 7 Error count for structure 2: det-Adj-N (Dulay & Burt, 1974a).

132

R. Azpilicueta-Martínez

(19) 1 *CHI1: In the floor [Additional: L1 preposition] (Spanish: ‘En el suelo’.) (20) 1 *CHI1: He is looking for the window to the rain [Additional: L1 syntax] (Spanish: ‘Él está mirando por la ventana la lluvia’.) (21) 1 *CHI1: There are two balloons flying? [Additional; subject-verb inversion: questions] (Spanish: ‘¿Hay dos globos volando?’) (22) 1 *CHI1: Balls of colours (used to mean ‘balloons’) [L1-structured circumlocution] (Spanish: ‘Bolas de colores’.) (23) 1 *CHI1: More long [Calque] (Spanish: ‘En un dibujo hay globos negros y los otros rojos’.)

5 Results and Discussion The present section provides an answer to the research questions above by dissecting explicit L1 use, L1 structural transfer, and by analysing the possible effect of the interlocutor and task variables in the students’ CLI. An additional subsection will cover observations of certain phenomena relevant to CLI in the learners’ production. Due to the small sample sizes, we resorted to the Wilcoxon signed-rank test (a nonparametric equivalent alternative to the matched-pairs t-test). Significance level was fixed at p ≤ .05. Quantitative analyses were carried out using SPSS Version 24.

5.1 L1 Use Regarding our first research question, explicit L1 use among the students in the present study (see Tables 4 and 5) was nearly inexistent, even lower than the—already low—rates in similar previous work on CLIL YLs (e.g., 4.8% for Year Three CLIL learners in García Mayo & Hidalgo, 2017) and in line with the low percentages reported in the EFL beginners in Lázaro-Ibarrola and Azpilicueta-Martínez (2015), that is, 0.51%. It is worth noting how, even at such low rates, students seem to diverge in the type of L1 they resort to depending on the interlocutor at hand, i.e., all of the L1 items uttered when interacting with an expert were forms of I:EE, that is, translation

Table 4 L1 use in children narrating to expert Day/Task D1T2 a I:EE

Utterances 494

INSERT: EXPLICIT ELICIT

L1 use 6

(I:EEa )

Percentage (%) 1.21

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

133

requests (Example 19), whereas most of the L1 terms displayed when interacting with a peer where mere (lexically meaningless) interjections of the ‘¡Ay!’ (English ‘Oh!’) type. (24) 1 *CHI1:… the girl is, erm… how do you say ‘quitar’? [INSERT: EXPLICIT ELICIT].

Evidently, this appears to relate to the fact that they might not regard their partners as a source of lexical knowledge in the same way as they do an adult pretend monolingual speaker of English. In this respect, even the I:EE raised in peer-peer interaction appeared to be also directed at the researcher, who was present while recording the interactions. This fact casts doubt as to whether participants would have resorted to the L1 explicitly or would have opted for other means of expression (e.g., circumlocution) to the same degree had the researcher not been present during their peer-peer interactions. The extremely low explicit L1 use rates in this work call into question the extent to which explicit L1 use necessarily declines as proficiency in the TL increases (Agustín-Llach, 2009; Herwig, 2001; Navés et al., 2005; Serra, 2007), since rates were still lower than the high-school students analysed in García Mayo and LázaroIbarrola (2015). As is the case with recent research in which older, more proficient CLIL groups resorted to their L1 more frequently (García Mayo & Hidalgo, 2017; García Mayo & Lázaro-Ibarrola, 2015), this study underscores the pivotal role of task features at cognitive and motivational levels and their subsequent impact on explicit L1 use.

5.2 L1 Structures With regard to our second and third research questions, instances of structural transfer through the students’ English were far more copious than explicit L1 use, as may be observed in the Tables 6, 7 and 8.

Table 5 L1 use in children narrating to a peer Day/Task

Utterances

L1 use

Percentage (%)

D2T1

203

3 (2I:EEa + 1 EDIT)

1.47

D2T2

329

7 (2 I:EEa + 5 EDIT)

2.12

TOTAL

532

10 (4 I:EEa + 6 EDIT)

1.87

a I:EE

INSERT: EXPLICIT ELICIT

134

R. Azpilicueta-Martínez

As may be observed in Tables 6 and 7 there was a higher rate of non-explicit CLI when students interacted with a peer. However, differences were non-significant (z = −1.218; p = 0.223). It is pertinent to highlight how D2T1 (Appendix) brought about the highest percentage of non-explicit CLI, more than doubling the rates in the other two tasks. A closer look at the data reveals how students appeared to struggle to convey the meaning of seemingly unclear differences, such as the ones in picture 1. Thus, these pictures might violate Criterion 28 (“Are actions clear?”) in Duff et al. (2008), and will be further considered in Sect. 5.2.1. We then compared the results of the students acting as narrators (N: 130) with the same students performing the task as story builders (N: 67), something which brought about significant differences (z = −2.361; p = 0.018). Nevertheless, we should not take these results at face value, since narrators nearly doubled the amount of output produced by story builders (1026 vs. 550 utterances). Finally, and even though it did not constitute the main focus of this study, possible differences regarding the interlocutor factor between the students performing the task as story builders (N: 27, D1; N: 40, D2) were also compared, again yielding non-significant results (z = −0.742; p = 0.458). Table 9 subcategorises the results regarding the classification in Sect. 4.4.2 when children acted as narrators (all tasks). In order to maximise TL use by the students, the usual response from the expert when asked to translate a given term was to pretend to misunderstand the speaker. This fact might have led to a decline in the number L1 items on the subjects’ part, and, possibly, to a higher rate of structural transfer given their need to look for alternative means of expression. Curiously, in one case did the researcher provide an answer to one of those questions, and it appeared to immediately trigger a higher number of translation requests—and, perhaps, to a lower effort to convey meaning using their own English words—within that dyad: (25) 1 *CHI1: In the… in the… how do you said… ‘carro’ (‘trolley’) (looking at researcher)? 2 *RES: Trolley? 3 *CHI1: In the trolley?

Table 6 L1 structures: children narrating to expert Day/Task

Utterances

Instances of non-explicit CLI

Percentage (%)

D1T2 (20 narrators)

494

47

9.51

Table 7 L1 structures: children narrating to children Day/Task

Utterances

Instances of non-explicit CLI

Percentage (%)

D2T1 (10 narrators)

203

46

22.66

D2T2 (10 narrators)

329

37

11.25

Total D2

532

83

15.60

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

135

Table 8 L1 structures: children narrating (total) Day/Task

Utterances

Instances of non-explicit CLI

Percentage (%)

Total

1026

130

12.67

Table 9 L1 structures: children narrating (total). Subcategories Structural transfer

Subcategory

N

Elision

Dummy subject

10

Percentage (%)

Preposition

21

16.2

Subject

14

10.8

7.69

Other forms: expanded

3

2.31

Plural forms

Double plurals

3

2.31

Overgeneralisation

Gender

Adjectives

Additional types of transfer

1 18

0.77 13.8

Particles ‘no’ and ‘yes’

2

Possessives

3

2.31

11

8.46

Subject-adjective inversion L1 prepositions L1 syntax (phrase level) Subject-verb inversion (questions) L1-structured circumlocution Calques

1.54

9

6.92

12

9.23

1

0.77

4 18

3.08 13.8

4 *CHI2: No. 5 *CHI1: Where? 6 *CHI2: How do you say ‘detrás’ (‘behind’)’? (looking at researcher) 7 *CHI1: Behind. 8 *CHI2: Behind the children. 9 *CHI1: Breads. 10 *CHI2: Ah… 11 *CHI1: In the third picture, in the fourth picture there, there are in the bookshop, the dad is reading a book and the two childrens are… erm… how do you say ‘agotados’ (‘drained’)? *RES: Researcher

As regards individual differences there was a great degree of variability among participants, including one subject producing no visible instances of L1 structural transfer whatsoever, while another single participant displayed 24 language tokens containing L1 structural transfer.

136

5.2.1

R. Azpilicueta-Martínez

Insights from the Students’ Interlanguage

In order to supplement the results in the previous subsections with more qualitative information, what follows next are observations from the participants’ production which were deemed of particular interest. One common form of non-explicit CLI among the students in D1T1 (see Appendix) was the following recurrent construction from different students interacting with the expert: (26) 1 *CHI1: That, erm… in the sofa, u, up the sofa are two balloons (27) 1 *CHI1: They are, there are balloons, erm… erm… erm…. erm… are more up of the sofa?

Students appear to share their spatial conceptualization leading them to resort to the same bespoke CLI alternative via ‘up’ or ‘up of’, which a monolingual speaker of English would have possibly discarded in favour of ‘over’. Further research would be needed in order to check the extent to which learners from different L1 backgrounds categorise and articulate this type of event in English. In Sect. 4.4.2 (iii) we have noted how the overgeneralisation of gender (‘boys’ including both genders in Spanish ‘chicos’) was present across the data. However, a closer look at this phenomenon reveals how task design appears to interact with lexical CLI particularly when word choice implies a significant difference between the pictures and the distractors. Example (28) was spotted in D1T2 (see Appendix), a story in which both children are visibly performing the same action in all pictures (i.e., it was unimportant to specify who did what): (28) 1 *CHI1: In the first picture… two boys are playing in the floor with the snow… and next there is a grandpa sitting in a bank

Example (29), by contrast, belongs to D2T1, a story in which the boy and the girl carried out different actions (upon which hinged some of the differences between the ‘right’ pictures and the distractors): (29) 1 *CHI1: Erm… the… boy and a girl are putting the… boots

Note how the child in 28 does not ‘bother’ to differentiate who is doing what, and so she produces a form of ungrammatical English with no undesirable consequences. D2T1, however, appears to ‘force’ students to be more accurate and avoid such overgeneralisation (see Sect. 4.4) if they are to succeed in the task. As a matter of fact, not a single student referred to ‘boys’ in any of the pictures in D2T1, and, when they did, they immediately self-corrected, as the example (30) below illustrates: (30) 1 *CHI1: Erm…. there are two boys in… there is a girl and a boy… seeing the bad day that… is in that moment

In their study, Lázaro-Ibarrola and Hidalgo (2017) had reported non-explicit CLI never leading to recasting or explicit corrections. While this phenomenon was also observed in the present work, occasionally confirmation checks or clarification requests were needed occasionally:

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

137

(31) 1 *CHI2: Yes. They are happy, they are sad? 2 *CHI1: Erm… sad? [Confirmation check]

Likewise, we believe the students’ sharing of the L1 might, in turn, be acting as a catalyst for the imitation of L1 structures reported by Lázaro-Ibarrola and Hidalgo (2017), since such imitation was also spotted right across our data. Note the absence of verb fronting in the questions in examples (31) and (32), belonging to the same dyad in two consecutive tasks: (32) 1 *CHI1: Student B: In the picture of the supermarket they are a… bread? Or, erm…. le…. Lettuce? 2 *CHI2: What? [Clarification request] 3 *CHI1: In the picture of the supermarket they are bread? Or… in… or, or… erm… or lettuce?

D1T2 (Appendix) yielded the lowest rate of non-explicit CLI, i.e., the highest percentage of grammatically correct utterances in the TL. We believe the results in D1T2 might be related to the fact that the story in it complied with key criteria in Duff et al. (2008) (such as clarity of the pictures and the logical sequence of the story) to a higher degree than the rest of tasks, and such clarity might have brought a less demanding, factual type of discourse on the students’ part. What follows are two typical samples of the participants’ production in D1T2: (33) 1 *CHI1: The girl is putting the eyes. 2 *RES: Very good! And? What happens after that? 3 *CHI1: They are thinking about and the old man is looking for her and is eating a banana. 4 *RES: Is the snowman finished? 5 *CHI1: Yes. 6 *RES: In picture number four? 7 *CHI1: No. 8 *RES: What’s missing? 9 *CHI1: The nose, and the scarf. 10 *RES: Oh, very good! 11 *CHI1: And the hat. 12 *RES: Ok! Excellent! How does the story finish, (student’s name)? 13 *CHI1: Then they do the snowman and the old man is very happy (34) 1 *CHI1: In the third the girl is putting the eyes in…to the snowman and the boys is putting the arms. 2 *RES: That’s very good, (student’s name)! What happens after that? 3 *CHI1: In the fourth picture there are, the children make the snowman and they are thinking what more they can put to the snowman and the father is… looking, eating the banana

Note how the structure and clarity of the story trigger a type of language among students which is simple, present-tense based, yet CLI-free to a high degree. Compare examples (33) and (34) to the following extract from D2T1 (35):

138

R. Azpilicueta-Martínez

(35) 1 *CHI1: Here two, two boys, one girl, one girl and two, and one boy, is raining and he, he looking of the window. 2 *CHI2: Erm… the… are two pictures… are two pictures the same. One, the… first picture it ha, the boy it’s sad and the girl is sad, and the second is with the… with the… with the mouth a little bit… erm… a little bit… straight. 3 *CHI1: Yes, is that

As opposed to examples (33) and (34), the case in 35 illustrates how the lack of clarity between the pictures, i.e., the violation of Duff et al.’s (2008) Criterion 28 (“Are actions clear?”) has generated the need in the story builder to negotiate with his interlocutor and attempt to clarify that two of his pictures were similar and it was hard for him to odd the wrong one out. In addition, while the distractor depicted a character in a mood beginner students are readily familiar with, ‘being sad’, the right picture included two children who could be expressing a range of more complicated, less clear mindsets, such as being baffled, scared, surprised or embarrassed. We believe this need to convey a more complex meaning might have impinged on the amount of non-explicit CLI in our participants. Interestingly, D2T2 included a slight violation of another of the criteria in Duff et al.’s (2008) Criterion 6: “Are the illustrations free of surreal or illogical elements?”. A significant portion of students took it for granted that picture 3 would be the one depicting the girl next to her daddy, since such was the position those characters appeared on the preceding and following scenes (as various students reported later). However, participants refraining from making that assumption tended to resort to either of the following two strategies: (36) 1 *CHI1: What, in what order are the children? First the boy or first the girl? 2 *CHI2: First the boy. Then they go to the library. The… children are bored, and the father is… is… see some a book (37) 1 *CHI1: And the boy is in the left or… in the right? 2 *CHI2: In the… In the right

Example (37) might appear closer to what a monolingual adult speaker of English would actually produce, since differences between those pictures hinge on a physical horizontal position, i.e., left-right, not ordinal or up-down. Surprisingly, example 36 proved to be much more effective in having the story builder choosing the right picture, whereas the more canonical, perhaps more CLI-free example (37) often led to confusion and the need to negotiate further again. Consequently, examples such as picture 3 in D2T2 might be promoting L1structured output rather than structures closer to the norm in the TL since such language choice might be (i) more efficient task-wise and also (ii) more similar to the students’ L1. In other words, the example above suggests the learner finds it both more effective and easier to use L1-structured TL than producing language closer to the TL standard. One possible explanation for the phenomenon above could be the possibility that children assumed ‘first’ would mean ‘first on the left’, the same pattern they follow when they read texts, while those opting for the second strategy (37) failed to use relative positions ‘left’ or ‘right’ efficiently, due to their lack of

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

139

fixed reference points (e.g., on the father’s right). This is confirmed in the following example (38) by another dyad: (38) 1 *CHI1: The girl is… at the… first or at the right? 2 *CHI2: Erm… the girl is… on the… right. 3 *CHI1: Fourth, please

It is worth noting how none of the participants resorted to prepositions of place in order to clarify picture 3 in D2T2, i.e., “Who is next to daddy?” or “Is the boy between daddy and the girl?”, even though they were familiar with them. As a whole, this subsection has attempted to supplement the results in Sect. 5.2 by providing qualitative insights on the intricacies and complexities of the students’ output, particularly in the ways task design (e.g., the violation of the criteria in Duff et al., 2008) interacts with CLI, and, in turn, the production in the TL.

6 Conclusions and Pedagogical Implications Results in this study have shown that explicit L1 use was extremely low, and values were similar irrespective of whether children interacted with a peer or an expert speaker of the TL. Thus, they concur with previous research questioning the inverted correlation between proficiency in the TL and explicit L1 use (Lázaro-Ibarrola & Azpilicueta-Martínez, 2015; Sánchez, 2003; Sanz, 2000). We believe these results can be accounted for by some of the factors mentioned in García Mayo and Hidalgo (2017), that is, “learner motivation, task complexity, task repetition and instructional setting” (García Mayo & Hidalgo, 2017, p. 133). Regarding motivation and explicit L1 use, the fact that the children in the study did not know the expert could speak Spanish8 might have led them to assume he would not understand any language other than English. If this were true, it might have impinged on their explicit use of the L1, and, consequently, might have triggered a higher use of roundabout expressions or L1 avoidance. This ‘perceived monolingualism’ within the interlocutor factor is a variable which ought to be explored in relation to L1 use and L1 CLI in future studies, given its profound pedagogical implications. Our results also appear to have implications regarding the ways task complexity, and the suitability and effectiveness of the visual prompts, affect explicit L1 use. The arrangement of the activity in dyads, and the students’ need to interact orally in order to arrange the story in the right order brought about no visible off-task communication, a type of interaction which had revealed a significantly higher number of L1 items in Broner (2001). Overall, our findings regarding explicit L1 use should push those teachers and stakeholders refraining from interactive activities with YLs on the grounds that these lead to less TL use. By contrast, non-explicit CLI appeared to be much more noticeable across the data. Structural transfer was consistently present irrespective of the task or interlocutor 8 Find

the raison d’être for this ‘pretended monolingualism’ on Sect. 5.2.

140

R. Azpilicueta-Martínez

at hand, therefore constituting a characteristic element of learners’ TL production at this beginner level. It is relevant to highlight that there were instances of the students’ interlanguage which correspond with the same processes in monolingual English-speaking children, such as the overgeneralisation of the negative particle ‘no’ (Schumann, 1979; Wode, 1978). On the other hand, our findings also warn about the possible fossilization of ungrammatical L1 structures in the TL when level-matched students interact in pairs. In this respect, some of the type and frequency of lexical CLI appears to be highly task-dependent, as noted in Sect. 5.2.1, naturally forcing the students to be more accurate in their language choice without the need of a pedagogical intervention. Results in the present work also led us to believe that task design, layout and complexity might also influence structural CLI. Some of the findings in Sect. 5.2.1 suggest features such as the degree of clarity or logic within a story might lead to the expression of more complex representations for which language in the students may be lacking and, in turn leading them to higher non-explicit CLI rates. In other words, when students share the L1, certain task features (e.g., the violation of some of the criteria in Duff et al., 2008) might lead them to resort to L1-structured TL rather than structures closer to the norm by being more effective task and communication-wise. We are, therefore, presented with the pedagogical challenge of designing and finding age, level and interest-appropriate tasks which boost students’ use of the TL (e.g., by pushing them into ‘on-task’ interaction) in paired interaction, that is avoiding tasks which might be too simple or not interesting enough, as García Mayo and Hidalgo (2017) forewarn with some of their older CLIL students. More research is needed however, in order to strike the right balance between tasks which are sufficiently interesting and cognitively-demanding for the students, yet not so ambitious that they prevent learners from using the TL structures needed to perform them successfully. As a result, findings in this study pave the way for further research analysing the possibility of a negative correlation between explicit L1 use and L1 structural transfer. As a whole, given the increasing questioning on the role of L1 in L2 learning regarding both teacher-learner and peer interaction (Cook, 2001; Turnbull & Arnett, 2002; Wells, 1999), the findings in the present study confirm that L1 terms are used scarcely and wisely in both interaction modes (child-child and child-expert), but warn us about the possibility that structural transfer could be reinforced when interaction occurs among peers at low levels of proficiency. Since such transfer might be triggered by learners’ perception of partial similarities between the L1 and the L2, and may be especially difficult to overcome when learners are frequently in contact with peers making the same errors (Lightbown & Spada, 2006), this paper advocates the use of task designs in which learners’ attention is also drawn to form, even at beginner level, and calls for further research on the developmental L2 readiness of learners at this level and age to benefit effectively from such practices.

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

Appendix: The Tasks D1T1: Birthday Party Story as seen by narrator, showing correct version:

1

2

3

4

5

141

142

R. Azpilicueta-Martínez

Story builder’s sheet, showing pictures in random order plus the following three distractors:

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

143

D1T2: The Snowman Story as seen by narrator, showing correct version:

1

2

3

4

5

144

R. Azpilicueta-Martínez

Story builder’s sheet, showing pictures in random order plus the following three distractors:

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

145

D2T1: On a Rainy Day Story as seen by narrator, showing correct version:

1

2

3

4

5

146

R. Azpilicueta-Martínez

Story builder’s sheet, showing pictures in random order plus the following three distractors:

What Lies Beneath: L1 Morphosyntax Seeping in Through Young …

147

D2T2: The Toyshop Story as seen by narrator, showing correct version:

1

2

3

4

5

148

R. Azpilicueta-Martínez

Story builder’s sheet, showing pictures in random order plus the following three distractors:

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL Context M. Luz Celaya, Laura Panelli and Júlia Barón

Abstract The present study aims at investigating the speech act of congratulating, a speech act that has received little attention to date, especially in foreign language learning settings with non-adult learners. More specifically, the study seeks to analyse the effects of age and proficiency and the influence of the pragmatics of the L1 in the production of this speech act by learners of EFL in a school context with no instruction on pragmatics. Participants were forty-nine 12 and 16-year-old learners of English (21 at grade 6 and 28 at grade 10, respectively) in a bilingual community in Spain. The instrument of data collection, which was specifically designed for the present study, was a WDCT with four different situations. Results show that younger (and less proficient) learners tend to use units that involve a focus on themselves rather than on the interlocutor whereas older learners, who are more proficient, use categories which directly involve the interlocutor, as expected in this speech act. Furthermore, the comparison of the learners’ congratulations in EFL to the productions in their L1s yields somewhat controversial results; on the one hand, there is evidence for the possible influence of the pragmatics of the L1 in the younger learners in some of the categories analysed, but there is also evidence for the effect of proficiency over L1 pragmatic influence in one of the categories. The study closes with a claim towards the need for future analysis in ILP in foreign learning contexts. Keywords ILP · EFL · Congratulations · L1 influence · Age · Proficiency

M. L. Celaya (B) University of Barcelona, Gran Via Corts Catalanes 585, 08007 Barcelona, Spain e-mail: [email protected] L. Panelli Council on International Educational Exchange in Barcelona (CIEE), Barcelona, Spain e-mail: [email protected] J. Barón Institute for Multilingualism at the Universitat Internacional de Catalunya, University of Barcelona, Barcelona, Spain e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_8

149

150

M. L. Celaya et al.

1 Introduction Interlanguage Pragmatics (henceforth ILP) has gained rapid interest within the field of applied linguistics in the last few decades. As has been claimed by Kasper and Rose (2002), among others, ILP can be considered as a ‘hybrid’ between interlanguage and pragmatics, and, thus, it analyses how learners of second languages acquire and develop the second language (L2) pragmatics. As defined by Bardovi-Harlig (2013, p. 68), ILP should be understood as “How-to-say-what-to-whom-when”. According to this, ILP carries both the more intrinsic pragmatic definition, as defined by Yule (1996, p. 3), “the study of the speaker meaning; the study of contextual meaning; the study of how more gets communicated than is said; the study of the expression of relative distance”, as well as the explanation of the acquisition of the L2 pragmatics, namely, the interplay between the pragmatics of the L1 and the L2 pragmatics itself. The unit typically used by researchers to analyse pragmatics is the so-called speech acts, defined by Yule (1996, p. 47) as the “actions performed via utterances”, a succinct definition which implicitly contains Searle’s (1976) components of a speaker, a hearer and an utterance. Speech acts have been analysed through different perspectives; in the case of ILP, the focus lies on “the comprehension and production of speech acts and how their L2-related speech act knowledge is acquired, as well as conversational management, discourse organization, or sociolinguistic aspects of language use such as choice of address terms” (Kasper & Dahl, 1991, p. 216). Several speech acts have been extensively analysed in ILP; however, congratulating has received little attention so far, especially in English as a foreign language (EFL) with non-adult learners. Therefore, the aim of the present study is to analyse the acquisition and use of this speech act by two groups of school learners with different ages and levels of proficiency in EFL with the main focus on the relationship between the pragmatics of the L1, age, and the level of proficiency in the L2, a factor that is intertwined with age in a school context, as claimed in Navés, Miralpeix, and Celaya (2005). The chapter has been organized as follows. First, we come across the definition of the speech act of congratulating in relation to several pragmatic issues (Sect. 2.1). There follows a review of studies on congratulating in Sect. 2.2. The review of the literature finishes with the presentation of the research questions. Section 3 explains the study itself (the piloting of the study, the context, the participants, the instruments employed, including the written discourse completion task (WDCT) and the personal questionnaire of the participants, the procedure, and, finally, the analysis of the data). Section 4 examines the results of the study, more specifically, the differences in types and frequency of congratulating units produced by the two age groups and the study of the possible influence of the L1 pragmatics. Section 5 contains the discussion of the results, and Sect. 6 addresses some pedagogical implications for the teaching of ILP in foreign language contexts; finally, Sect. 7 presents the conclusions and limitations of the present study.

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

151

2 Review of the Literature 2.1 Defining the Speech Act of Congratulating Congratulating was included in both Austin’s (1962) and Searle’s (1976) categorization of types of speech acts. Austin (1962) classified congratulating in the category of behabitives, which implies the intention of the interlocutor to express his own attitude towards the other person’s behaviour. Searle (1976) further categorized the speech act of congratulating as an illocutionary speech act, which is the type where the interlocutor’s intention is to express his own psychological perception of a situation, presumably shaped by an intention of being polite. As Searle (1976) claimed, illocutionary acts help all human beings to express their feelings, interests, needs and attitudes through them, and they usually express more than one of these ideas together. Illocutionary acts were also sub-divided into: assertives, directives, commissives, expressives and declaratives. According to this categorization, congratulating can be considered as belonging to expressives, since such type of speech act expresses a psychological state towards a proposition. Following Brown and Levinson’s (1987) concept of face, the speech act of congratulating constitutes one of the representations of face-enhancement politeness; face contains those positive or negative aspects which are universal to all human beings and are susceptible of being threatened or enhanced by some specific utterances. In this line, and according to Leech (1983), congratulations are said to have a convivial function that are intrinsically courteous and, consequently, closely linked to the use of politeness. More recently, however, such perspective on politeness has been criticised in Spencer-Oatey’s (2005, p. 336) rapport management theory on the belief that politeness “is concerned with the (dis)harmony in social relations”. In a previous study, Spencer-Oatey (2002) questions the association between face and politeness, and advocates for the inclusion of a social identity component to counter balance the individual concept of negative face. Thus, the researcher (2002, p. 540) distinguishes between “quality face” (“face is associated with personal/social value, and is concerned with people’s sense of worth, credibility, dignity, honour, reputation, competence and so on) and “social identity face” (“concerned with the value that we effectively claim for ourselves in terms of social or group roles”). Therefore, in this perspective the focus moves from the linguistic strategies that are used to the evaluation of those strategies by both speakers and hearers (in this line, see García, 2009) for the analysis of congratulations and rapport management by Peruvian Spanish speakers). By drawing on such previous ideas, and as mentioned in the introduction, ILP aims at understanding how the L2 pragmatics is acquired and developed. Kasper and Schmidt (1996, p. 154) stated that, even though certain pragmatic strategies are universally available for certain speech acts, that is, “conveying pragmatic intent indirectly and making use of routine formulae”, there are differences that should be learnt by the non-native speakers (NNS) in order to communicate successfully with a native interlocutor in an L2. That is, despite the fact of being universal and

152

M. L. Celaya et al.

easily found in all cultures, politeness is also shaped by the different socio-cultural values present in each speech community (see, for instance, Blum-Kulka, House, & Kasper, 1989). In relation to the languages involved in the present study, it has been claimed that English is a negative-politeness oriented language whereas Spanish is positive-politeness oriented (see Ballesteros Martín, 2001; Peláez Torres, 2017).

2.2 The Speech Act of Congratulating: A Review of Studies As claimed by Bardovi-Harlig (2013), more research on acquisitional pragmatics is needed, even if several developmental studies have been carried out (such as Achiba, 2003; Celaya & Barón, 2015; Rose, 2000, among many others). This is even more evident in relation to the speech act of congratulating, as will be seen below, although other expressive speech acts such as complimenting, greetings and thanking have received some attention so far as in Placencia and Lower (2013), Taponen (2014), De Pablos Ortega (2011), and Hickey (2005), respectively. The scant research on the speech act of congratulating mostly involves culturally distant first and second languages. Thus, Elwood (2004) carried out a cross-cultural analysis of responses to a WDCT of 45 American students writing in English, 45 Japanese students writing in English and 45 Japanese students writing in Japanese. The participants answered to seven situations dealing with happy news. Elwood found different patterns depending on the group; the answers provided by Japanese students writing in English or in Japanese were usually accompanied by comments of envy or self-related comments whereas American students writing in English usually asked for more information related with the happy news. Depending on the situation, the most frequent answer provided varied. For the situation that represented a promotion, the most commonly used strategy was the Illocutionary Force Indicating Device (IFID), whereas in the situation recreating a wedding, participants usually chose expression of happiness. With the taxonomy proposed by Elwood (2004), Allami and Nekouzadeh (2011) also conducted a study on congratulating strategies. Fifty Iranian Persian speakers completed a discourse completion test (DCT) with nine situations presenting happy events but differentiated by degree of social distance with the interlocutor (intimate, acquaintance, and stranger). Results from the study revealed that the most commonly used strategy to congratulate was IFID, indicating that Iranian speakers mainly made use of exaggerations, giving gift to listeners, or in-group identity markers when congratulating others. Another relevant study is the one conducted by Dastjerdi and Nasri (2013), who investigated cross-cultural differences in the use of the speech act of congratulating by American, Persian and Syrian Arabic native speakers. Dastjerdi used a WDCT including 4 different situations presenting good news. The findings showed that the American participants’ most frequently used formulas were IFID, request for information and offer of good wishes, whereas the other two groups of participants presented similarities in the use of strategies that, despite their low frequency, were

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

153

only found among them, as asking the interlocutor for sweets or making use of humorous statements. In a similar vein, Nasri, Dastjerdi, and Ghadiri (2013) carried out a study with Persian, English, and Armenian native speakers who completed a WDCT with four natural situations. The American participants were found to use the same formulas as in the study above (mainly IFID, request for information and offer of good wishes). Persian and Armenian participants also used IFID and offer of good wishes but Armenians also used expression of happiness while Persians speakers used request for information. Bataineh (2014) carried out a study on the different politeness strategies adopted by 50 Jordanian native speakers of Arabic and 50 American native speakers of English on the speech acts of congratulating, thanking, and apologizing, by using a WDCT with 9 items representing happy events. In this case, contrary to the studies reviewed above, results showed that the two groups used the same strategies, regardless of their cultural differences. This study supported the idea of linguistic universality, as both groups selected strategies that resulted in friction-free, goodwill, and promotion of the cooperation. Similar conclusions are reached by Reyes (2011) in her study of the speech act of congratulation in different languages and cultures. The participants were 42 intermediate and advanced learners of L2 Mexican Spanish who reflected upon congratulating in their first languages. The instrument used presented several situations where the participants had to answer if they would congratulate people on those specific occasions in their first languages. An interesting feature of the instrument is the division into situations that imply an effort on the part of the potential recipient of the congratulation and those which do not. After the analysis of the answers, Reyes claims that congratulating is similar across cultures, although she also acknowledges the low number of participants from some of the countries as a drawback of the study and, at the same time, the need for further research (for instance, eight of the countries/languages had only one participant each). Finally, Panelli (2014) investigated the speech act of congratulating in grade 6 and grade 10 EFL Spanish learners. The participants in her study answered a WDCT where they had to congratulate their interlocutor in four different situations involving social distance between speakers; two of the situations were designed to involve a high degree of familiarity and two a lower degree. An interesting finding appears in the differences between groups in terms of degree of familiarity. Grade 6 participants did not show differences in their responses based on social distance with the interlocutors, whereas grade 10 learners produced more responses involving their personal opinions and interest in situations with a close social distance between interlocutors. The above review reveals that studies on congratulating are still scarce in our context with Spanish/Catalan bilingual learners of EFL. The present study is, then, an attempt to fill this gap by analysing the speech act of congratulating. In the light of the findings reviewed above, the present study aims at analysing how two groups of Spanish/Catalan learners of EFL use the pragmatics of the target language; more specifically, it addresses the possible influence of the L1 pragmatics in relation to age and proficiency in the foreign language. For this, two different groups were selected, one group from grade 6 of primary school, and another group from grade 10 of secondary education. The following research questions are thus proposed:

154

M. L. Celaya et al.

1. Is there any difference in the speech act of congratulating produced by grade 6 and grade 10 EFL learners regarding the types and frequency of congratulating units in their productions? 2. Does the speech act of congratulating show any influence from the L1 pragmatics in these groups of learners?

3 The Study 3.1 Pilot Study Before data collection, the research instrument was piloted to test its adequacy and to assess the feasibility of the study. The instrument was first discussed with the primary school teacher whose pupils would become the grade 6 group participants. She proposed two changes, namely, the inclusion of several illustrations to support the description of the situation, and the simplification of the language in the instructions. The second version of the instrument was revised and given to three students at grade 6 and four students at grade 10, who were selected at random from other classes and excluded from the final sample. Results from the pilot study confirmed that the instrument was adequate for both levels of proficiency and ages.

3.2 Context The sample included students from two different state-run schools in Barcelona where English was taught as the only foreign language. Both schools are situated in similar neighbourhoods, two areas in Barcelona with low socio-economical background. Teachers for both groups are native Spanish speakers with a degree in English Philology and many years of experience in teaching EFL. Personal interviews with the teachers revealed that the methodology followed in the two courses is the communicative approach and, in both cases, English is the language of instruction.

3.3 Participants The initial sample of participants for the study included 54 learners; 24 participants were from an intact class from the primary school and 30 were from an intact class from the secondary school. However, data from the sociocultural questionnaire (see below) showed that five of them did not meet the required criteria, so the final sample included 49 learners, all of them bilingual in Spanish and Catalan. Criteria for including them in the study were the following: the students could not have

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

155

attended any extra-curricular English lessons, they could not have been on a stayabroad in an English-speaking country or have English as their native language, they could not know any other foreign language besides English, and, finally, they should have started instruction in English at the age of 6. The number of hours of instruction of the participants was equal for all of them in each group, that is, a total of approximately 288 h of instruction received by grade 6 (G6 henceforth) students and a total of approximately 432 h received by grade 10 (G10 henceforth) students. The final sample thus included 49 participants (21 at G6, with a mean age of 11.6 and 28 at G10, with a mean age of 16.4). As stated in the introduction, age and proficiency level are intertwined in foreign language learning in a school context. Additionally, data were gathered in Spanish or Catalan from a total of 26 participants (12 at Grade 6 and 14 at Grade 10) as a baseline data of productions in L1 Spanish and Catalan. These participants were not the same learners who carried out the task in English to avoid the effects of task repetition of the instrument they had answered in English; however, they attended schools with similar features as those described above.

3.4 Instruments The instrument used in this study was a WDCT. The reason for using this instrument was that, as Bardovi-Harlig (2013) pointed out, this type of task allows researchers to control as many variables as possible (especially, topic, participants and contexts), and so make findings comparable with results from similar types of tasks. We are aware of the fact, however, that DCTs, whether oral or written, are not without criticisms on the grounds that the use of such instruments should be triangulated with other types of instruments and data to ensure validity, as claimed by EconomidouKogetsidis (2013). The WDCT in the present study was adapted from Allami and Nekouzadeh (2011) and from Panelli (2014) (see Sect. 3.6). As previous studies have done, for instance, Koike and Palmiere (2011), the situations were designed considering real life situations; hence, although the whole task is in written format, two of the scenarios involve written interaction and two involve a simulation of face-to-face interaction. The interlocutor’s gender was alternatively assigned so there are two male and two female speakers in the situations. Furthermore, the test included pictures, which according to Bardovi-Harlig (2013), increases naturalness, and the four situations involve topics related to the learners’ ages and background (see Appendix for the WCDT). A linguistic and socioeconomic background questionnaire was used with both primary and secondary learners. The questionnaire contained eight items related to their experience with the target language; one of the questions asked on their knowledge of other languages besides their native ones. This information was used

156

M. L. Celaya et al.

to include only participants with exposure limited to English at school and knowledge of no other second languages, as explained above.

3.5 Procedure First, data were collected from primary-school participants by the second author of the present study, with the presence of their teacher. The researcher explained the purpose of the task, highlighting the fact that there was no right or wrong answer for the situations. Every situation was explained in detail and participants could ask any doubts they had in relation to the task. The maximum time allowed to complete the task was 15 min. Right after this, participants filled in the personal questionnaire. Students were asked to sign anonymously or with a pseudonym. A week after this, data were collected in the secondary school following the same procedure. Finally, data in L1 Spanish and Catalan were gathered from the 26 participants, as described in Sect. 3.3.

3.6 Analysis The total number of situations analysed in the present study is 196 (four from each of the 49 participants). Data were encoded and analysed following a scale of congratulating categories adapted from Allami and Nekouzadeh (2011) and from Panelli (2014). First, data were categorized using Allami and Nekouzadeh’s scale, considering that each response from each participant in every situation may contain more than one structure of the same type. After categorizing the data, it was observed that certain strategies presented in the scale by Allami and Nekouzadeh were not found in the participants’ responses of this study and, at the same time, certain strategies used by these participants did not have a matching category in the model proposed by these researchers. For this reason, the final scale did not include all the items in Allami and Nekouzadeh’ s and, on the other hand, a few items were created ad hoc; similarly, some of Panelli’s (2014) items were grouped into the same category and one of them was rejected because it was considered irrelevant for the aims of the present study. The resulting instrument of analysis consisted in eight categories which imply different ways of congratulating (see Table 1); an important aspect to bear in mind regarding this categorization has to do with the issue of whether the focus of the speech act lies on the person who provides the congratulation unit or on the person who receives the actual congratulation. Interrater reliability was carried out between the three researchers of the present study and with an anthropologist who was used to analysing data qualitatively. The agreement reached was 98 and 97.75%, respectively. Then, data were analysed with SPSS (version 22.0). Both descriptive and inferential statistics (Mann-Whitney u test) were performed.

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

157

Table 1 Scale adapted from Allami and Nekouzadeh (2011) and from Panelli (2014) 1. Illocutionary force indicating device (IFID): “congratulations”. Expressions of praise: “you are the best”. Offer of good wishes or hopes: “good luck”. Expressions stating the good luck of the interlocutor “You are very lucky” 2. An expression assessing the situation positively: “That’s great” 3. Statements of personal happiness: “I’m so glad” 4. Self-Related comments. Expressions of envy or bad feelings: “I’m jealous” 5. Expressions of personal opinion: “I like Berlin”. Personal requests: “Can you invite me?” Exclamations/Expressions of surprise: “I can’t believe it” 6. Affection expressions: “Honey” 7. Suggestions to celebrate or to meet together: “let’s go out for dinner” 8. Statements suggesting actions: “Visit Barcelona as well”. Encouragements: “now you have to study more” Unedited examples from the data of the present study

4 Results The classification of the tokens into eight categories of congratulating units, as specified above, allowed us to address the first research question, which inquired into the possible differences between G6 and G10 in both the frequency and types of categories in the production of the speech act of congratulating. As shown in Table 2, the descriptive analysis shows that G6 and G10 differ in the types of categories that are produced the most. Thus, G6, the younger and less proficient group, presents the highest number of tokens (74) in category 2 (expression assessing the situation positively), as in “Great”, which is followed by category 5, e.g. “I go to you, plis?” (expressions of personal opinion, personal requests and exclamations/expressions of surprise) and by category 3, e.g. “I happy for you” (statements of personal happiness) with 46 and 45 tokens, respectively. Category 1, which consists of units that indicated IFID, expressions of praise, offer of good wishes or hopes and expressions stating the good luck of the interlocutor, as in “Congratuleisons” presented 31 tokens. Category 4, which was characterized by those expressions which showed self-related comments and expressions of envy or bad feelings was produced 24 times, e.g. “I envy”. The categories that present the fewest tokens are category 6 (affection expressions), as in “Kiss for you” and category 7, as in “We could Paris!” (suggestions to celebrate or to meet together), with only 7 and 2 tokens, respectively. The type of unit belonging to category 8 is not produced by any of the G6 learners. G10 participants, on the other hand, present quite a different order in the categories that are most frequently used. They produce the highest number of tokens (106) in category 1, e.g. “Congratulations, you got it!” and “You are a very good student and your academy future are very good”, which is followed at a large distance by category 2, as in “It’s fantastic”, with 62 tokens. There follow category 5 (36 tokens), e.g. “Can you invite me, please? I want to go!”, category 3 (33 tokens), e.g. “You will see your favourite band and I’m very happy” and category 8, those expressions which

158

M. L. Celaya et al.

suggested actions and encouragements, as in “You have to celebrate it with your family and friends”, with 24 tokens. The categories that present the lowest number of tokens are categories 6 with 13 tokens, e.g. “What’s app bro?” and categories 4 and 7, with 10 and 6 tokens, respectively, as in “I’m jealous but it’s happy” and “We couldn’t talk for such a long time, I hope that now, we could talk some more”. The Mann-Whitney u test applied to the data revealed statistically significant differences in category 1 (z = −4.240, p = .000), in which G10 students produced more tokens of such congratulating strategy (M = 3.79, Std = 1.988, N = 28) than those students from G6 (M = 1.48, Std = 1.123, N = 21). In line with these results, category 8 also showed significant differences (z = −3.924, p = .000), and even if it was not one of the most frequently preferred categories, it was produced by G10 students (M = .86, Std = 1.079, N = 28), whereas G6 learners did not produce any tokens of this category at all, as indicated above (M = .00, Std = .000, N = 21). Statistically significant differences between the production of G6 and G10 were also found in categories 3, 4 and 5. In these cases, however, it was G6 participants who presented a higher number of tokens than G10 learners. Category 3 (z = −2.880, p = .004) is produced by G6 students (M = 2.14, Std = 1.062, N = 21) more than by G10 learners (M = 1.18, Std = 1.278, N = 28). Similar findings were also reported in category 4 (z = −3.655, p = .000) and 5 (z = −3.223, p = .001), for which G6 learners showed a higher number of items in such category (M = 1.10, Std = .889, N = 21) than those produced by G10 (M = .25, Std = .499, N = 28). Similarly, category 5 was preferred by G6 learners (M = 2.38, Std = 1.359, n = 21) over G10 learners (M = 1.04, Std = 1.232, N = 28). However, no significant differences were found in categories 2, 6 and 7 (z = −1.928, p = .054, z = −1.066, p = .286 and z = −.849, p = .396, respectively). Descriptive statistics showed that both G6 and G10 learners seemed to produce a similar number of congratulating strategies in category 2: (M = 3.62, Std = 2.519, N = 21) (Md = 2.21, Std = 1.881, N = 28). The results suggest that categories 6 and 7 seemed to be the least preferred ones for both G6 (M = .33, Std = .966, N = 21—in the case of category 6—M = .10, Std = .301, N = 21, in the case of category 7), and G10 (M = .46, Std = .838, N = 21, in category 6; and M = .28, Std = 1.499, N = 28). Table 3 presents a summary of the results from the statistical analysis. In sum, results show that for those categories that yielded a statistically significant difference, G10 participants, that is the older and more proficient learners, presented more tokens belonging to categories 1 and 8 than G6 learners, but fewer tokens belonging to categories 3, 4 and 5.

Table 2 Raw number of tokens per group for each of the categories

Cat 1

Cat 2

Cat 3

Cat 4

Cat 5

Cat 6

Cat 7

Cat 8

G6

31

74

45

24

46

7

2

0

G10

106

62

33

10

36

13

6

24

Total

137

136

78

31

82

20

8

24

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

159

Table 3 Results from the Mann-Whitney u test Categories

p

Mean

Standard deviation

G6

G10

G6

G10

N G6

G10

Category 1

.000

1.48

3.79

1.123

1.988

21

28

Category 2

.054

3.62

2.21

2.519

1.881

21

28

Category 3

.004

2.14

1.18

1.062

1.278

21

28

Category 4

.000

1.10

.25

.889

.499

21

28

Category 5

.005

2.38

1.04

1.359

1.232

21

28

Category 6

.286

.33

.46

.966

.838

21

28

Category 7

.396

.10

.21

.301

1.499

21

28

Category 8

.000

.00

.86

.000

1.079

21

28

The second research question focused on L1 pragmatic influence in both groups of learners in the production of the speech act of congratulating. To answer this question, the frequencies and types of tokens produced by both groups were analysed in relation to the production in L1 Spanish/Catalan by the same age speakers collected as base line data (see Sects. 3.3 and 3.5). Both G6 and G10 behaved similarly in their L1 in two ways; first, they produced a higher number of tokens belonging to category 1 and category 8 than the participants who produced these categories in EFL, and, secondly, their productions in their first languages were more elaborate and consisted of longer units than those analysed in the L2 data. In fact, as seen in relation to the first research question, category 8 was the only type that G6 learners did not produce at all in the L2. Contrary to this, two of the categories (2 and 4) were produced either sporadically or not at all in the L1. Category 4 was not used by any of the two age groups in their L1 whereas, as seen above, G6 learners produced tokens belonging to this category in the L2 and so did G10 learners, although to a much lesser extent. Category 2 presents very few tokens in either group; however, it is one of the preferred categories in the L2. Finally, a relevant difference between the two age groups when congratulating in their L1 appears in category 5, more specifically when it is produced to focus on oneself and not so much on the interlocutor, as in the case of requests such as “Can you invite me?”. Younger L1 speakers (G6) used such requests as a congratulating speech act extensively, whereas older speakers (G10) did not; the same findings appear in the L2, as can be seen above.

5 Discussion The first research question aimed at examining whether there were any differences in the speech act of congratulating by G6 and G10 learners, that is, younger and less proficient learners as compared to older and more proficient learners. The results

160

M. L. Celaya et al.

described above show that there actually seem to be differences both in the frequency and the types of units when used by learners at different ages and proficiency levels. Although, as mentioned above, it is difficult to separate age and proficiency in the present study, a qualitative analysis of the tokens together with the quantitative analysis of the data seem to point to two trends as to the weight of one or the other factor in the interpretation of the results. To begin with, the effect of proficiency can be seen when considering the categories most frequently produced by G6 and G10 learners. G6 students preferred using those units from categories 2, 3 and 5, and differences with G10 were statistically significant for categories 3 and 5. The units from such categories are characterised mainly by fixed expressions, so it could be argued that they are structures which tend to be memorised instead of being analysed. This would go in line with previous ILP studies on other speech acts, which have found that fixed expressions tend to be first acquired in pragmatic development (see Achiba, 2003; Barón & Celaya, 2010; Celaya & Barón, 2015), so they are commonly used by low proficiency learners. Such claim also seems to apply to the findings from G10, who also preferred using units from categories 2 and 5, thus rejecting this possible effect of proficiency at first sight. In a similar vein, although for the speech act of apologies, Sabaté i Dalmau and Curell i Gotor (2007) concluded that a high level of proficiency in the L2 does not always imply being more pragmatically fluent. It is important to notice, though, that G10 learners presented the most number of tokens in category 1. Even if the congratulations from such category may also be fixed expressions (such as ‘congratulations’ or ‘good luck’), category 1 is characterised by the use of more creative language and longer units and even the incorporation of other speech acts, such as suggestions or encouragements. This fact may imply that the more proficient learners become, as in G10, the more complex structures they use (see Félix-Brasdefer, 2007), but that fixed expressions are still used as a way of promoting fluency. In the study by Barón and Celaya (2010), more proficient learners still produced fixed expressions but they also presented a wider variety of request units, as well as routines and patterns, than lower proficiency learners and younger learners. We suggest that this seems to also be the pattern to describe our G6 and G10 learners in relation to the effect of proficiency. The effect of L2 proficiency is more evident when analysing category 8, which G6 learners did not produce at all, contrary to G10 learners who have more linguistic tools at their disposal. Let us recall here that category 8 yielded statistically significant differences between both groups. Moreover, the base line data in L1 also allows us to confirm such claim, since category 8, together with category 1, is widely used by G6 and G10 speakers, who produce long elaborate units because they do not have to struggle with the foreign language. The second trend comes to light when we bear in mind that units from categories 3 and 5 are generally more self-centred than the units used in the other categories (see the scale used for the present study). It could be argued, then, that the preference of younger learners, those from G6, for this type of units is related to age, since in terms of cognitive maturity they do not seem to be acknowledging the interlocutor’s

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

161

position (see Panelli, 2014). However, as they grow older, they might become more socio-pragmatically aware, so they can also put themselves on the interlocutor’s perspective. Such finding seems to point to a possible effect of age on the speech act of congratulating. The second research question aimed at analysing whether the L1 pragmatics was related to the L2 learners’ production of the speech act of congratulating in EFL. To this aim L1 and L2 data were compared with the assumption that the production of similar types of congratulating units in L1 and in L2 by each of the groups could imply a possible influence of the L1 pragmatics or the existence of similarities in congratulating strategies among different cultures, as proposed in Reyes (2011), and also in line with the findings in Bataineh (2014). Categories 3, 4 and 5 are those that congratulate the interlocutor by focusing on oneself, an approach that has been associated with a young age. The Spanish/Catalan speakers at G6 used categories 3 and 5 to congratulate in their L1 whereas older (G10) speakers did not. This was also the behaviour found in the production in L2, since both categories 3 and 5 yielded statistically significant differences with higher means in G6. Our interpretation of these findings is that L1 pragmatic influence and age appear in an inverse relationship in L2 production, that is, the younger the learners, the more they draw on the pragmatics of their L1. However, the opposite finding also appears in the analysis of the data. As was pointed out in the Results Section (Sect. 4), when considering the L1 data, it was found that both G6 and G10 learners used congratulating units from category 8; in L2 production, on the contrary, only G10 learners produced it. Since category 8 implies the use of more elaborate language, this difference between G6 and G10 would suggest that it is proficiency and not L1 pragmatic influence that is at stake when producing such category. Since the students from both groups are already competent in their L1, they can use units from category 8 in their L1, but only the more proficient ones are able to do so in the L2. The comparison of the production of category 4 (self-related comments, expressions of envy or bad feelings) in L1 and in L2 is an interesting case in point since we expected to come across it in congratulations in L1 Spanish/Catalan by the younger group, as seen in the L2, where the category yielded statistically significant differences between the two groups. However, since no L1 informants produced units in such category, we may claim that learners did not rely on their L1 pragmatics. The explanation as to why units such as ‘I’m jealous’ are produced by G6 learners in EFL may lie in the fact that they could have learnt them as chunks and used them to be more fluent. This is in line with Elwood (2004), who found expressions of envy in situations in which happy events were presented, although in that case the languages and corresponding cultures involved in the study were far apart.

162

M. L. Celaya et al.

6 Pedagogical Implications We believe that the findings in the present study have some relevant pedagogical implications in relation to ILP in foreign language classrooms. To begin with, we have seen that the choice of some of the students’ pragmatic expressions was somehow triggered by their age. Accordingly, learners in FL contexts might benefit from being taught the types of pragmatic moves that correspond to those used at different age ranges. Such approach implies the use of appropriate situations (and materials) to deal with speech acts, that is, situations that directly appeal to the specific age, and not only to the level of proficiency of the learners. In addition to this, since L1 influence in ILP seems to be more evident in younger learners, teachers should also take this fact into account when teaching young learners so as to provide them with metapragmatic explanations which would not only help them in their L2 pragmatic development and gradual acquisition of a richer pragmatic repertoire in the L2, but would help them as well to gain more pragmatic awareness both in the L1 and L2. In this sense, we suggest the use of situations that imitate those in their daily interactions in the L1, an approach which, at the same time, goes hand in hand with the cultural component present in ILP. Finally, although the speech act of congratulating in the present study has been analysed in learners with no explicit instruction on pragmatics, several studies have proved the benefits of being instructed in the L2 pragmatics, as stated in the Discussion Section (Sect. 5). As has been claimed by Alcón (2008), EFL contexts do not usually include pragmatics as part of the syllabus. Therefore, pragmatic development may be a difficult task for L2 learners in such contexts (Barón & Celaya, 2010). However, as some studies have shown, instruction seems to be beneficial for learners, because it helps them in pragmatic development. The fact that the participants in the present study did not receive instruction on pragmatics might have been one of the reasons for the little elaboration of units used in their productions. If learners had been instructed, they might have used a wider variety of tokens and intensifiers when congratulating. This would go in line with those studies which have explored the effects of teaching on the speech act of congratulating in EFL contexts. Thus, in Eslami-Rasekh and Mardani (2010) the groups which received instruction outperformed those who did not receive pragmatic instruction, especially in the use of intensifiers; in Bardovi-Harlig and Vellenga (2012), it was claimed that instruction was beneficial and directly related to levels of proficiency. The lack of studies on the speech act of congratulating may suggest that there is a need to carry out more research on the nature of such speech act in EFL learners in order to know how it can be addressed in the foreign language class at different ages and proficiency levels. Therefore, we think that future studies should explore the speech act of congratulating both from a developmental and an instructional perspective.

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

163

7 Conclusion The present study aimed to shed some light on the acquisition of the speech act of congratulating by EFL learners. First of all, the purpose of this study was to explore whether there are differences in the production of both types and frequencies of congratulating units between younger and less proficient learners (Grade 6 learners) and older and more proficient learners (Grade 10 learners). The findings suggest that some differences could be identified between the two groups. On the one hand, younger and lower proficiency learners (G6) used units which were characterized by being more formulaic in nature than those produced by older and more proficient learners (G10). Even if fixed expressions were also used by G10 learners, they also included more complex congratulations. On the other hand, G6 learners seemed to prefer those units which were more self-centred whereas G10 learners used units that directly involve the interlocutor, as expected in such speech act. The second aim of this study was to examine whether the learners’ L1 pragmatics had an influence on their L2 production. The findings suggest that the pragmatics of the L1 have an effect on the L2 productions in younger learners. However, proficiency seems to outweigh the effect of L1 pragmatic influence for some of the categories analysed. The present study has three main limitations. First, the low number of participants in the study; a larger sample would lead to more generalizable results. Second, even if two of the situations in the instrument used simulate oral interaction, the instrument itself is a written task; spoken interaction would also provide more information about the development of the speech act of congratulating. Finally, the fact that, as stated above, age and proficiency usually go together when analysing learners in a school context is a confound that may make the interpretation of some of the results inconclusive. Notwithstanding, it is hoped that the present study might have contributed to a better understanding of the acquisition and use of the speech act of congratulating in an EFL school context, since ILP is still a much unexplored area of research in foreign language settings. Acknowledgements We are grateful to the teachers and students who participated in the study, and very especially to Silvia López, the primary school teacher with whom we piloted the WDCT, and who suggested several important ideas to adapt the test to the specific groups in the study. We want to express our gratitude to the two anonymous reviewers for their insightful feedback, which has undoubtedly contributed to the improvement of the chapter. Any errors remain our own.

164

M. L. Celaya et al.

Appendix: WDCT Situation 1: Graduation. One of your Facebook friends is from New York. You met her last summer in Barcelona but she is not a close friend of yours. Her Facebook status says that this academic year has been really difficult for her, but she finally is getting excellent in all the courses. You are happy for her and. Send her a private message on Facebook. Your friend’s Facebook status: I made it! After hours and hours of studying day and night, without going out, without internet, without Facebook or Twitter, without videogames (sigh)… I made it! I passed all courses! I can’t believe I’m graduating!

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

165

Situation 2: Getting a prize. There is a British student from London studying at your school. He is staying just a semester and he is not a close friend of yours but you usually talk. Today he tells you that he just won two tickets for his favourite band’s concert in Berlin. You know is the only concert his favourite band is giving in Europe. He looks really happy and excited and he tells you:

Your friend: I won the tickets! I couldn’t imagine I could win! I’m flying to Berlin to see my favourite band in concert! Oh my God, I can’t believe it!!

166

M. L. Celaya et al.

Situation 3: Enrolling on a football school. One of your best friend lives in Manchester. You are very good friends and you keep your contact via email. He wrote you an email telling you that he has been selected to be part of the Manchester United Soccer School. He is really excited about that. You know that he loves playing football and you are also truly happy for him. You send him an email.

Your friend’s email: Hi! How is everything going in Barcelona? Here things are going great because I’ve been selected to be part of the Manchester United Soccer School! This school is one of the best in Manchester! I’m starting this summer. I can’t wait to go there! And, who knows, maybe I will become the next best football player in the world! I am so happy!! Answer me and tell me what you think! Best, A.

“I’m Jealous but I Am Very Happy”: Congratulating in an EFL …

167

Situation 4: Travelling all over Europe. One of your best friends is from California. You met her during her stay in Barcelona. You are close friends despite the distance. Today you are talking via Skype and she tells you that her parents promised her to visit Europe this summer as a birthday present. She loves Europe and this is very important for her. She is really happy and, of course, you are happy for her as well.

Your friend: You know what? I celebrated my birthday party last Saturday and my parents promised me that this summer we all could travel around Europe! We could visit Paris! I still can’t believe that present! I am so happy!!

The Use of Communication Strategies in L3 English CLIL Learners Izaskun Arratibel-Irazusta and María Martínez-Adrián

Abstract Most of the investigations on communication strategies have been carried out in English as a Foreign Language (EFL) contexts. More limited research exists in Content and Language Integrated Learning (CLIL), and has mainly focused on the use of previously known languages. Thus, there is a need of studies that tackle other strategies such as target language (TL)-based strategies. This paper analyses the use of TL-based strategies and previous language-based strategies in an oral narration task and its correlation with general proficiency and knowledge of receptive vocabulary in two different age/proficiency CLIL groups of third language (L3) English learners. Results indicate that proficiency does not have a big impact on the use of these strategies. As regards the distribution of strategies, holistic strategies (i.e., TL-based strategies) are the most used, a finding ascribed to the overriding effect of CLIL. With respect to correlation analyses, although previously known language-based strategies do seem to align with the level of proficiency and receptive vocabulary, this does not apply to the use of TL-based strategies. Additionally, the use of TL-based strategies does not imply a lower use of prior linguistic experience, which is still relevant in these CLIL learners to scaffold L3 production. Keywords CLIL · Communication strategies · Previously known languages · Target language-based strategies

I. Arratibel-Irazusta (B) · M. Martínez-Adrián Department of English and German Philology and Translation and Interpreting, Faculty of Arts, University of the Basque Country UPV/EHU, Unibertsitateko Ibilbidea 5, 01006 Vitoria-Gasteiz, Spain e-mail: [email protected] M. Martínez-Adrián e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_9

169

170

I. Arratibel-Irazusta and M. Martínez-Adrián

1 Introduction The use of communication strategies during oral production has been the focus of attention in numerous studies in the Second Language Acquisition (SLA) literature (Cenoz, 2001, 2003a; Lázaro Ibarrola & García Mayo, 2012; Martínez-Adrián & Gutiérrez-Mangado, 2015; Muñoz, 2007; Poulisse, Bongaerts, & Kellerman, 1990), and in particular is currently receiving increasing attention in the area of Content and Language Integrated Learning (CLIL), in which the present study is framed. Communication strategies refer to all those devices language learners use to overcome linguistic difficulties encountered when trying to communicate in a foreign language with a reduced interlanguage system (Fernández Dobao, 2002). Among the existing classifications of communication strategies, three of them are the most popular in the area: Tarone’s taxonomy (1977), Færch and Kasper’s taxonomy (1983) and the Nijmegen group’s taxonomy (Poulisse, 1990). In the case of CLIL research, studies on communication strategies have specifically taken into account the classifications by Tarone (1977) and Poulisse (1990). Figure 1 summarizes the different categories considered by these two taxonomies. In the present chapter, in line with Tarone’s (1977) taxonomy, appeals for assistance (‘What is this? What called’ from Tarone, 1977, p. 199), also known as interactional strategies by other authors (Cenoz, 2001, 2003a), will be considered.1 Conceptual and transfer strategies will also be investigated following Poulisse’s taxonomy (1990). Conceptual communication strategies (also called target language (TL)-based strategies) encompass both holistic and analytic strategies. Holistic strategies are “the use of a superordinate, coordinate, or subordinate term” such as the use of ‘vegeta-

COMMUNICATION STRATEGIES

Fig. 1 Types of communication strategies

1 In

particular, those appeals or interactional strategies produced in L1 Spanish/L2 Basque will be codified.

The Use of Communication Strategies in L3 English CLIL Learners

171

bles’ for peas (Poulisse, 1990, p. 59). Analytic strategies, on the other hand, refer to circumlocution, description, and paraphrase such as ‘this you use for a baby so, uh, that it can’t uh, make, uh, his clothes erm uh dirty’ (Poulisse, 1990, p. 59). Transfer is broken down into borrowings (Dutch ‘etalage’ for English ‘shop-window’), foreignizings (‘cuffer’ from French ‘coiffeur’) (Poulisse, 1990, p. 60) and literal translations (‘my elders’ for English ‘parents’) (Poulisse, 1990, p. 109), but only the former (borrowings and foreignizings) will be explored in this paper. The use of mime (as clapping one’s hands to illustrate applause) (Tarone, 1977, p. 199) and avoidance (topic avoidance or message abandonment) (Tarone, 1977, p. 199) will not be tackled in the present study. Several factors have been found to affect the use of communication strategies such as: proficiency level in the TL (i.e., Bialystok, 1983; Jourdain, 2000; Liskin Gasparro, 1996; Paribakht, 1985; Poulisse et al., 1990; Tarone, 1977); gender (Jiménez Catalán, 2003; Wang, 2008); task-related features like cognitive demands, time constraints and interlocutor’s role (Khanji, 1993; Poulisse et al., 1990), among others. Most investigations have been carried out with English as a foreign language (EFL) learners (Cenoz, 2003a; Gost & Celaya, 2005; Muñoz, 2007; Poulisse & Bongaerts, 1994; Viladot & Celaya, 2007). More recent studies have examined CLIL learners. However, research is still in its infancy in this respect and the vast majority of these studies have focused on previously known language-based strategies (i.e., Arratibel-Irazusta & Martínez-Adrián, 2018; Lázaro Ibarrola & García Mayo, 2012; Martínez-Adrián, in press; Martínez-Adrián & Gutiérrez-Mangado, 2015), with the exception of Martínez-Adrián, Gallardo-del-Puerto, and Basterrechea (2019), who have examined the whole inventory of strategies as depicted in Fig. 1. Thus, there is a need of studies that tackle other communication strategies such as TL-based strategies (i.e., the use of analytic and holistic strategies) together with previously known language-based strategies. Furthermore, most of the studies that have explored the use of communication strategies in the CLIL literature have compared CLIL to non-CLIL learners (i.e., Celaya & Ruiz de Zarobe, 2010; Martínez-Adrián & Gutiérrez-Mangado, 2015) and few (pseudo)longitudinal studies have been conducted to the present date (i.e., Lázaro Ibarrola, 2016; Lázaro Ibarrola & García Mayo, 2012). Additionally, to the knowledge of the authors there is a lack of studies addressing a correlation between the use of communication strategies (previously known language-based strategies and TL-based strategies), general proficiency, as well as receptive vocabulary measures. According to several studies conducted in CLIL and non-CLIL learners, general proficiency correlates with receptive vocabulary knowledge (Iglesias Diéguez & Martínez-Adrián, 2017; Nemati, 2010; Quián, 1999). Similarly, CLIL slightly increases the correlation between receptive vocabulary and general proficiency, suggesting that a bigger size of receptive vocabulary relates to a more successful performance in general proficiency (Iglesias Diéguez & Martínez-Adrián, 2017). As a higher level of proficiency entails a lower use of previous language-based strategies (Martínez-Adrián & Gutiérrez-Mangado, 2015) and given the relationship between general proficiency and receptive vocabulary knowledge attested in these previous investigations, an increase in receptive vocabulary

172

I. Arratibel-Irazusta and M. Martínez-Adrián

knowledge should presumably lead to a lower use of previously known languagebased strategies and a greater use of TL-based strategies. This paper will try to fill these gaps by examining the use of previously known languages (L1 Spanish/L2 Basque) in appeals for assistance and transfer (borrowings and foreignizings) together with TL-based strategies in an oral narration task in two different age/proficiency CLIL groups of L3 (third language)2 English learners. It will also correlate the findings obtained in this respect to the results from a general proficiency level test and a receptive vocabulary level test. This chapter is organized in seven sections. Section 2 reviews the main empirical findings with respect to the use of communication strategies in CLIL and non-CLIL settings. In the third section, the research questions are addressed while the study is described in the fourth section. Sections 5 and 6 present and discuss the results obtained in the study. The main conclusions from the study are depicted in Sect. 7. The chapter finishes with a set of pedagogical implications in Sect. 8.

2 Research on Communication Strategies Communication strategies have been analyzed in several studies which have tried to shed more light on the classification, effectiveness, teachability and factors affecting choice of communication strategies (Martínez-Adrián et al., 2019). Among the factors considered in these investigations, proficiency is the factor that has received the greatest attention in research. In terms of amount, less proficient learners tend to rely more on communication strategies than their more proficient learners due to their limited knowledge of the TL (Hyde, 1982; Liskin Gasparro, 1996; Poulisse et al., 1990). However, results are still inconclusive with respect to the choice of particular types of communication strategies. Although some authors (Bialystok, 1983; Jourdain, 2000; Wannaruk, 2003) have concluded that less proficient learners tend to use more prior linguistic knowledge-based strategies, avoidance and mime, while more proficient learners prefer TL-based strategies (i.e., paraphrases), the impact of proficiency has been found to be overruled by other factors such as the nature of the communicative task (Fernández Dobao, 2002; Poulisse et al., 1990). For example, in Fernández Dobao (2002), advanced learners used as many avoidance and transfer strategies as the low-proficient learners did in highly-demanding tasks. In the Spanish context, in which the present investigation was conducted, studies have mainly analyzed the use of previously known languages in interactional strategies (appeals for assistance, clarification requests and metacomments) and transfer (i.e., borrowings, foreignizings and calques), instead of the whole inventory of strategies (i.e., Martínez-Adrián et al., 2019). The use of these strategies has been widely studied in non-CLIL contexts (Celaya & Torras, 2001; Cenoz, 2001, 2003a; 2 The term L3 will be used throughout the paper to refer to English, the third language known by the

participants from the present study, who speak Spanish, which counts as one language, and Basque, which counts as a second language.

The Use of Communication Strategies in L3 English CLIL Learners

173

Muñoz, 2007; Navés, Miralpeix, & Celaya, 2005). More limited research exists in the case of CLIL settings. Some of these studies compare CLIL learners to nonCLIL learners as regards the use of previously known languages in oral and written production (i.e., Agustín-Llach, 2009; Celaya, 2008; Celaya & Ruiz de Zarobe, 2010; Gallardo-del-Puerto, 2015; García Mayo & Lázaro Ibarrola, 2015a; Lázaro Ibarrola, 2016; Martínez-Adrián, in press; Martínez-Adrián & Gutiérrez-Mangado, 2015; Pladevall Ballester & Vraciu, 2017). Other studies are (pseudo)longitudinal studies of CLIL learners but are lower in number (i.e., Arratibel-Irazusta & MartínezAdrián, 2018; Lázaro Ibarrola & García Mayo, 2012; Martínez-Adrián et al., 2019). In general, these studies have reported a lower use of previously known languages by CLIL learners and have confirmed a negative correlation between general proficiency, lexical richness and use of previously known languages (i.e., Martínez-Adrián & Gutiérrez-Mangado, 2015). However, even if CLIL learners use their previously known languages to a lower extent, some functions are still performed in these languages in these meaning-oriented programs (Lo & Lin, 2019; Martínez-Adrián, in press). More specifically, in terms of interactional strategies, CLIL learners do not rely so much on their previously known languages, a finding ascribed to a higher proficiency level usually attained by these learners thanks to the exposure to more intense and meaningful input provided in CLIL lessons (García Mayo & Lázaro Ibarrola, 2015a; Martínez-Adrián, in press; Martínez-Adrián & Gutiérrez-Mangado, 2015). The examination of the results in some of these investigations has also revealed greater differences between CLIL and non-CLIL learners as grade increases which could be explained by the effect of the accumulated hours of CLIL instruction received (Martínez-Adrián, in press). As for the distribution of previously-known languages in interactional strategies (appeals for assistance, clarification requests and metacomments), CLIL learners are more used to speaking in the TL and to interacting because of the active methodology they follow in the classroom (Coyle, 2007), which could explain the lower use of the L1 in the category ‘metacomments’ usually attested in some of these studies that have examined interactional strategies (i.e., Martínez-Adrián, in press). In addition, we cannot forget that the integration of content and language in content-based instruction is normally achieved by means of content tasks in the foreign language that are cognitively engaging for the learners (García Mayo, 2015), such as the ones carried out by participants in some of the aforementioned studies. So, it may be the case that CLIL learners are better equipped to perform such tasks. Similarly, metacomments are usually produced in the previously known languages in CLIL and non-CLIL learners (Martínez-Adrián, in press). This is consistent with data coming from studies that offer observation data from the CLIL classroom (Gené Gil, Garau, & Salazar Noguera, 2012; Pastrana Izquierdo, 2010). The use of prior linguistic experience is common in unplanned discourse in CLIL, which has led some authors to demand a reconceptualization of classroom code-switching as something which is natural for bilinguals to do (Lin, 2015; Swain & Lapkin, 2013). In fact, the ability to use multiple languages is regarded as part of the communicative competence of multilinguals (Arnfast & Jorgensen, 2003; as cited in Lo & Lin, 2019).

174

I. Arratibel-Irazusta and M. Martínez-Adrián

Additionally, teachers in different meaning-oriented classrooms have been observed to use prior linguistic experience to different extents and for different purposes so as to bridge the wide gap existing between students’ L2 proficiency and cognitive levels (Coyle, Hood, & Marsh, 2010). But despite the general consensus among researchers and teachers in CLIL programs about the benefits of using prior linguistic knowledge for both content and language learning in these contexts (Lo & Lin, 2019), many of them advocate a principled and judicious use of these languages (Lasagabaster, 2013). As regards transfer (borrowings, foreignizings and calques), while CLIL learners produce fewer borrowings in oral (Gallardo-del-Puerto, 2015; Pladevall Ballester & Vraciu, 2017) and written (Agustín-Llach, 2009; Celaya & Ruiz de Zarobe, 2010) production, results in the case of foreignizings are inconclusive. Whereas some studies analyzing written production have observed a greater production of foreignizings in CLIL learners due to their greater proficiency (Agustín-Llach, 2009; Celaya, 2008; Celaya & Ruiz de Zarobe, 2010), more recent research examining oral data has proved the opposite (Arratibel-Irazusta & Martínez-Adrián, 2018; Gallardo-delPuerto, 2015; Martínez-Adrián et al., 2019). However, as mentioned in Gallardodel-Puerto (2015), the studies that have attested a higher number of foreignizings in CLIL learners due to their higher proficiency are not supported by statistical evidence. Therefore, the production of foreignizings cannot be due to a higher proficiency. In the case of calques, CLIL learners have been found to resort to them to a higher extent (Gallardo-del-Puerto, 2015). In contrast to these studies that have focused on previously known languages as a communication strategy, to the knowledge of the authors, just one study conducted in a CLIL setting has been done on the use of previous linguistic experience-based strategies together with other strategies such as mime, avoidance, TL-based strategies and morphological creativity (i.e., Martínez-Adrián et al., 2019). This study examines CLIL learners’ self-reported opinions about the use of communication strategies in grades 5 and 6 of primary education. The analysis of the data reported similarities between grades 5 and 6 which were explained in the light of the small gap in proficiency between both groups of learners. Likewise, the most preferred strategies were appeals for assistance and paraphrasing and the least, morphological creativity. In this respect, CLIL was suggested to overrule the effect of proficiency as the beginner learners immersed in this type of meaning-oriented classrooms tended to favor the use of TL-based strategies, which are more typical of learners with a higher proficiency, rather than avoidance and previously known language-based strategies, which typically characterize low-proficient learners. The meaningful input learners are exposed to in a content-based language teaching context seems to favor their use of more advanced communication strategies. Teachers in these input-rich contexts tend to paraphrase more frequently (Wood, Bruner, & Ross, 1976) so as to avoid communication breakdowns and the learners in this study might be mimicking this behavior. Furthermore, the fact that the TL in these CLIL settings is used as a means of communication might lead them to be able to manipulate concepts in the TL to a larger extent. Nevertheless, even if these learners reported using strategies more typical of advanced learners, they manifested an extensive use of appeals for assistance,

The Use of Communication Strategies in L3 English CLIL Learners

175

a finding in line with previous studies conducted with young learners in a foreign language setting (Victori & Tragant, 2003). Even though the review of the studies done in this section reveals a growing body of research in CLIL contexts in Spain, (pseudo)longitudinal studies which could shed more light on the development of strategy use are still very limited in number (i.e., Arratibel-Irazusta & Martínez-Adrián, 2018; Lázaro Ibarrola, 2016; Lázaro Ibarrola & García Mayo, 2012; Pladevall Ballester & Vraciu, 2017). Additionally, there is scarcity of research in CLIL settings on the use of other communication strategies different from prior linguistic experience-based strategies (i.e., MartínezAdrián et al., 2019). The results obtained in the present investigation can contribute to the existing debate regarding the coexistence of previously known languages and the TL in the CLIL class to develop students’ bilingualism and biliteracy (Tedick & Weseley, 2015). There is also a lack of research addressing a correlation between the use of communication strategies, general proficiency, as well as vocabulary measures (i.e., Martínez-Adrián & Gutiérrez-Mangado, 2015), two areas particularly enhanced by CLIL. Thus, this article will try to fill these gaps by investigating the use of TL-based strategies and previously known language-based (Basque and Spanish) strategies in two different age/proficiency CLIL groups of L3 English learners.

3 Research Questions Based on previous findings regarding the use of communication strategies in CLIL settings, we address the following research questions: RQ. 1: Are there any differences between more and less proficient learners with respect to the use of previously known language-based and TL-based strategies? RQ. 2: Which is the predominant communication strategy in both groups? RQ. 3: Are there any correlations between general proficiency, receptive vocabulary, and the use of communication strategies (i.e., previously known language-based strategies and TL-based strategies)?

4 The Study 4.1 Participants A total of 48 students participated in the study. These students are Basque/Spanish bilinguals, with a very similar sociolinguistic status, learning L3 English in a semiprivate school in the Basque Country. All of them come from Spanish-speaking families as this is the majority language in the area where these learners live. However, all of them are enrolled at school in the bilingual B model which is mainly intended for native speakers of Spanish who want to be bilingual in Spanish and

176

I. Arratibel-Irazusta and M. Martínez-Adrián

Table 1 Participant characteristics Age at testing

Age of first exposure

Years of exposure

Total number of hours of exposure

CLIL I

13–14

CLIL II

15–16

3

10

1887

3

12

2057

Basque. As Lasagabaster (2007) mentions, this model could be considered as an early partial immersion program in which more or less half of the subjects are taught through Spanish and the other half through Basque, with the exception of EFL that is instructed in English. Students who are enrolled in this model obtain a high command in L2 Basque so they are considered Spanish/Basque bilinguals dominant in L1 Spanish (Cenoz, 2009). The context in which the subjects are immersed has been defined as additive trilingualism, where the language of instruction (Basque) is a minority language, Spanish is the majority language, and English is the main foreign language (Cenoz & Valencia, 1994). All the participants started learning English at the age of three for an hour a week and were enrolled in a CLIL program at the age of six, in which they were taught Social Sciences in English for half an hour a week. Participants were divided into two different groups. The first group (CLIL I) was composed of 28 participants (11 females and 17 males), whereas the second (CLIL II) was made up of 20 students (12 females and 8 males). CLIL I participants were 13–14 year-old students in the 2nd year of secondary education. At the moment of testing, they had received approximately 1887 h of English instruction, including English language classes and CLIL classes. CLIL II participants, on the other hand, were 15–16 year-old students who were in the 4th year of secondary education and at the moment of testing, they had received approximately 2057h of English instruction, encompassing English language classes and CLIL classes. Participant characteristics are summarized in Table 1. The younger participants of the study, even if they have been exposed to the TL fewer years than the older learners due to their younger age, they have been exposed to a higher number of hours per year from their 4th year of primary education onwards as a result of the incorporation of more content subjects in English in the curriculum. In the 4th year of primary education, apart from the English language and the Social Sciences classes, the younger participants have received Music lessons for an hour a week. In addition, since the introduction of English language classes at the age of three and the introduction of CLIL classes at the age of six, all the subjects taught in English have been incorporating more hours per week as students in both groups advance from one course to the following one. Thus, at the time of testing, the youngest group, who was in the 2nd year of secondary education, was receiving 3h per week of EFL lessons, 3h per week of Social Sciences and 2h per week of Music. In contrast, the oldest group was receiving 3h per week of EFL lessons and 3h per week of Social Sciences. Thus, at the time of testing, the youngest group was

The Use of Communication Strategies in L3 English CLIL Learners

177

receiving 2h of additional instruction in English per week. In addition, all students in both groups received on average 2h per week of extra exposure by means of private classes.

4.2 Instruments and Data Analysis Data were gathered by means of three different instruments. Participants were first tested on general proficiency by means of the standardized Oxford Quick Placement Test (QPT), Version 1 (UCLES, 2001). Secondly, the 2k Vocabulary Level Test (VLT) (Schmitt, Schmitt, & Clapham, 2001) was administered in order to measure their knowledge of receptive vocabulary. Finally, all of them were asked to narrate the well-known story ‘Frog, where are you?’ (Mayer, 1969) to the researcher in order to analyze their use of communication strategies. All the narrations were audiotaped and later on transcribed into the CHILDES format (McWhinney, 2000). As for the codification, the QPT consisted of a total of 60 items assessing reading, vocabulary and grammar and they were all in multiple-choice format, that is, for each item students had to choose the correct answer out of four different options. For each correct answer one point was given so that the maximum score participants could get was 60 points. In the 2k VLT, students had to match a target word with the corresponding definition. A total of 60 target words were used for testing. However, only 30 definitions were provided. The test was divided into ten groups of six words and three definitions. For each correct answer, that is, for matching each definition with the correct word, one point was given, so that the maximum score participants could get was 30 points. In order to calculate the total number of words known by participants, Nation’s formula (1990) was applied, according to which learners’ vocabulary size is obtained by calculating the total number of correct answers in the vocabulary test multiplied by the given band (2000 in this case), and then divided by the total number of items in the test (30 items). For the oral narration task, all cases of use of previously known language basedand TL-based strategies were identified. Given the status of both Spanish and Basque in the school context, we looked for cases of L1 Spanish/L2 Basque. The use of previously known language-based strategies were broken down into appeals for assistance and transfer. On the other hand, instances of TL-based strategies were subdivided into analytic and holistic strategies in line with Poulisse (1990).

4.2.1

Previously Known Language-Based Strategies

Appeals for Assistance Appeals for assistance refer to those cases in which the learner asks the interlocutor for help so as to produce a specific term in English.

178

I. Arratibel-Irazusta and M. Martínez-Adrián

(i) ¿cómo se dice how say ‘How do you say rock?’

roca? rock

(ii) agujero? hole ‘Hole?’

Transfer Transfer refers to the use of one or more terms (but not whole sentences) in Basque or Spanish as part of an utterance produced in English (see also Cenoz, 2003a in this respect). It encompasses categories such as borrowings, foreignizings and literal translations (Poulisse, 1990). However, in the present study only foreignizings and borrowings will be analyzed. (iii) It have his room desordenate it have his room messy ‘His room was a mess’ (iv) In the tree was a one buho. in the tree was a one owl ‘In the tree, there was an owl’

4.2.2

TL-Based Strategies

Analytic Strategies Analytic strategies allow the speaker to refer to an intended concept by listing (some of) its properties (i.e., circumlocution, description, and paraphrase) (Poulisse, 1990). In the present study, all these strategies will be codified as a whole, that is, no distinction will be made between circumlocution, description, and paraphrase. (v) The house ‘Beehive’

of the

bees.

Holistic Strategies Holistic strategies allow the speaker to refer to a concept by using the word for a related concept (i.e., the use of a superordinate, coordinate, or subordinate term) (Poulisse, 1990). As in the case of analytic strategies, all these strategies will be arranged as a whole. (vi) ‘Hamster’ for ‘mole’.

The Use of Communication Strategies in L3 English CLIL Learners

179

With respect to statistical analyses, results were analyzed by means of SPSS 19 (IBM Corp., 2010) to investigate the existence of statistically significant differences between the two groups, and to find out significant correlations between communication strategies, general proficiency and receptive vocabulary measures. In order to test whether the data was normally distributed, the skewness and kurtosis values were first calculated (Larson-Hall, 2010). For the cases in which the data met the criteria for normal distribution, t-tests were conducted for between-group comparisons. On the contrary, when the data did not achieve the criteria for normal distribution, Mann-Whitney u tests were run. For intra-group comparisons, Wilcoxon’s Signed Rank Tests and Friedman Tests were performed. Finally, for correlation analyses, a Spearman’s Rho correlation test was used.

5 Results First, results concerning the QPT will be shown. Then, results regarding the VLT will be reported. Subsequently, the oral narration task will be analyzed in order to examine communication strategies (both prior language-based and TL-based strategies). Finally, correlation analyses will be carried out so as to determine whether there are any significant correlations between the two tests administered and the use of communication strategies in the story telling task.

5.1 Quick Placement Test Table 2 presents the results of the Quick Placement Test. The mean score for the CLIL I group was 26.07 out of 60, whereas for the CLIL II group, the mean score was 33.00 out of 60. Thus, the CLIL II group of older learners was found to have a higher proficiency in the TL when compared to their younger counterparts. As the sample is normally distributed, a t-test (Table 2) was conducted in order to check whether the differences between the mean scores of the two CLIL groups were statistically significant. The test revealed a statistically significant difference in favor of the CLIL II group (t = −3.997; p-value = 0.001)

Table 2 Results from the Quick Placement Test GRADE

N

Mean

Minimum

Maximum

SD

CLIL I

28

CLIL II

20

26.07

13

37

6.13

33.00

24

45

5.77

180

I. Arratibel-Irazusta and M. Martínez-Adrián

Table 3 Results from the Receptive Vocabulary Test GRADE

N

Score

Words

Maximum

SD

CLIL I

28

17.71

1181.20

Minimum 6

26

5.99

CLIL II

20

24.35

1623.60

19

29

3.13

5.2 Vocabulary Level Test Results from the receptive vocabulary test and the statistical analysis appear in Table 3. As can be observed, the mean score for the CLIL I group was 17.71 out of 30, whereas for the CLIL II group, the mean score was 24.35 out of 30. Students’ scores were transformed into the number of known words applying Nation’s formula: ‘Vocabulary size = N correct answers multiplied by total N words in dictionary (the relevant word list) divided by N items in test’ (Nation, 1990, p. 78). The mean number of words obtained for the CLIL I group was 1181.20, whereas for the CLIL II group, the mean number of words was 1623.60. As expected, older learners performed better in the test, that is, their knowledge of receptive vocabulary was bigger than their younger counterparts. It is also important to mention that in both groups, the mean score is higher than 1000 words, which indicates that they know more than the most frequent 1000 words in the TL. As the sample is approximately normally distributed, a t-test was conducted in order to check whether the differences between the mean scores of the two CLIL groups were statistically significant. The difference between the groups turned out to be highly significant (t = −4.98; p-value = 0.001).

5.3 Oral Production Task In this section, results regarding the overall number of communication strategies and the total amount of words produced in both groups will be provided in the first place. Then, results regarding previously known language-based strategies will be shown. Subsequently, the results with respect to TL-based strategies will be presented. Finally, a comparison of the different categories in both groups will be offered. Table 4 depicts the results regarding the ratio of communication strategies per total number of words produced for each group. The numerator reflects the number of communication strategies and the denominator the total number of words produced for each group. As can be observed, the CLIL I group produced longer narratives and made use of communication strategies to a higher extent than the CLIL II group.

Table 4 Ratio of communication strategies per word Ratio of communication strategies

CLIL I

CLIL II

3.39%

2.26%

299/8819

136/6028

The Use of Communication Strategies in L3 English CLIL Learners

181

Table 5 L1 use in interactional strategies CLIL I

CLIL II

Mean

Median

Min.

Max.

SD

Mean

Median

Min.

Max.

SD

2.93

1.00

0

12

3.58

1.95

0.00

0

12

3.63

Table 6 L1 use in transfer lapses CLIL I

CLIL II

Mean

Median

Min.

Max.

SD

Mean

Median

Min.

Max.

SD

1.79

1.00

0

13

3.02

0.55

0.00

0

3

0.89

5.3.1

Previously Known Language-Based Strategies3

Appeals for Assistance Table 5 shows the results obtained in both groups for interactional strategies. By looking at Table 5, it can be observed that younger students produced more instances of appeals for assistance than their older and more proficient counterparts. In order to test whether this difference between groups was statistically significant, a Mann-Whitney u test was run. The test revealed the inexistence of statistically significant differences (z = −1.121; p-value = 0.262).

Transfer In Table 6, results with respect to transfer are summarized. As can be observed, the CLIL I group produced more instances of transfer when compared to their older and more proficient counterparts. In order to check whether these differences were statistically significant, a Mann-Whitney u test was conducted for between-group analysis in which a statistical tendency was found (z = −1.830; p-value = 0.067).

5.3.2

TL-Based Strategies

Analytic Strategies The results regarding the use of analytic strategies are displayed in Table 7. As can be observed, the CLIL I group seemed to use this strategy more frequently than the CLIL II group. However, a Mann-Whitney u test revealed the inexistence of significant differences between the groups in this respect (z = −0.412; p-value = 0.680). 3 No

instances of Basque were found in any of the groups, so results reflect the use of L1 Spanish appeals for assistance and transfer.

182

I. Arratibel-Irazusta and M. Martínez-Adrián

Table 7 Use of analytic conceptual strategies CLIL I

CLIL II

Mean

Median

Min.

Max.

SD

Mean

Median

Min.

Max.

SD

1.71

1.00

0

6

1.82

1.25

1.00

0

4

1.07

Table 8 Use of holistic conceptual strategies CLIL I

CLIL II

Mean

Median

Min.

Max.

SD

Mean

Median

Min.

Max.

SD

4.25

3.00

0

17

4.14

3.05

3.00

1

7

1.70

Holistic Strategies Results with respect to holistic strategies are illustrated in Table 8. As in previous cases, younger participants seemed to use this strategy more frequently than older learners. However, the Mann-Whitney u test used did not yield a statistical significance (z = −0.464; p-value = 0.642).

5.3.3

Comparison of the Different Categories

Table 9 presents the comparison of the different communication strategy categories in the CLIL I group. As can be observed, holistic strategies are the most frequent type of strategy followed by appeals for assistance. On the contrary, analytic strategies are the least used strategy followed by transfer. In order to see whether these differences were statistically significant a Friedman’s Test was conducted. Statistically significant differences emerged (Chi-Square = 8.271; p-value = 0.041). Consequently, post-hoc analyses were run. In order to do so, a Wilcoxon’s Signed Rank Test was used. Results showed that holistic strategies differed significantly from transfer but not from appeals for assistance. On the other hand, no significant differences were found between analytic strategies and previously known language-based strategies (appeals for assistance and transfer). Within the category of previously known language-based strategies, no significant differences were found between appeals for assistance and transfer.

Table 9 Instances of communication strategies in the CLIL I group CLIL I Previously known language-based strategies

TL-based strategies

Appeals

Analytic

Transfer

Holistic

Mean

SD

Mean

SD

Mean

SD

Mean

SD

2.93

3.579

1.79

3.023

1.71

1.823

4.25

4.142

The Use of Communication Strategies in L3 English CLIL Learners

183

Table 10 Wilcoxon’s Signed Rank Test for the CLIL I group

z

Transferappeals

Analyticappeals

Holisticappeals

Analytictransfer

Holistictransfer

Holisticanalytic

−1.433

−1.075

−1.134

−0.578

−2.898

−2.772

0.152

0.282

0.257

0.563

0.004

0.006

Sig. (2-tailed)

Table 11 Instances of communication strategies in the CLIL II group CLIL II Previously known language-Based strategies

TL-based strategies

Interactional Strategies

Transfer

Analytic

Mean

SD

Mean

SD

Mean

SD

Mean

SD

1.95

3.634

0.55

0.887

1.25

1.070

3.05

1.701

Holistic

Table 12 Wilcoxon’s Signed Rank Test for the CLIL II group

z Sig. (2-tailed)

Transferappeals

Analyticappeals

Holisticappeals

Analytictransfer

Holistictransfer

Holisticanalytic

−1.642

−0.209

−1.819

−2.287

−3.604

−3.232

0.101

0.834

0.069

0.022

0.001

0.001

Within the category of TL-based strategies, a significant difference was found in favor of holistic strategies. Table 10 shows a summary of the interactions. For the comparison in the CLIL II group, the same descriptive analysis was carried out. Table 11 shows the results. As in the CLIL I group, holistic strategies are the most frequently used strategy, followed by appeals for assistance. In contrast, transfer is the category that was least frequently used followed by analytic strategies. A Friedman’s Test revealed statistically significant differences between the categories (Chi-square = 22.278; p-value = 0.001). Thus, as in the CLIL I group, post-hoc analyses were conducted. A Wilcoxon’s Signed Rank Test showed that, in this group, holistic strategies differed significantly from transfer and appeals for assistance. In addition, analytic strategies also presented a significant difference with respect to transfer. Within TL-based strategies, holistic strategies significantly differed with respect to analytic strategies. Within previously known language-based strategies, the analysis did not yield a statistically significant difference between transfer and appeals for assistance. Table 12 presents a summary of the interactions.

184

I. Arratibel-Irazusta and M. Martínez-Adrián

Table 13 Results from the Spearman’s Rho correlation test CLIL I Spearman’s Rho QPT & VLT

CLIL II Sig. (2-tailed)

Spearman’s Rho

Sig. (2-tailed)

0.691

0.001

0.698

0.001

QPT & L1-based

−0.501

0.007

−0.411

0.072

VLT & L1-based

−0.555

0.002

−0.252

0.283

QPT & TL-based

0.058

0.770

−0.232

0.324

VLT & TL-based

−0.141

0.473

−0.085

0.722

L1 & TL-based

−0.023

0.909

−0.140

0.556

5.4 Correlation Analyses In order to test whether there were any significant correlations between the two tests administered as well as the use of previously known language-based strategies (appeals for assistance and transfer) and TL-based strategies (analytic and holistic), a Spearman’s Rho correlation analysis was conducted for each of the groups. As previously mentioned, due to the lack of influence from Basque, only L1 Spanish instances were included in the correlation test. Table 13 presents the results. As can be observed in Table 13, there are some statistically significant correlations (marked in bold). For both groups the QPT and the VLT showed a significant positive correlation in both groups, which means that the higher the general proficiency, the higher the receptive vocabulary. As regards the QPT and L1-based strategies, there was a significant negative correlation in the CLIL I group, and a statistical tendency was found in the CLIL II group. As for the VLT and L1-based strategies, a significant negative correlation was found in the CLIL I group as well. That is, in the younger group, the higher the receptive vocabulary, the less L1-based strategies. However, this does not apply to the older group as no significant correlations were established for these categories. In addition, no significant correlations emerged for any of the groups between TL-based strategies and the tests measuring general proficiency and receptive vocabulary. No significant correlations were found between the use of L1-based strategies and TL-based strategies either.

6 Discussion In this section, we will answer the three research questions for the present study. With respect to our first research question (Are there any differences between more and less proficient learners with respect to the use of previously known language-based and TL-based strategies?), just a statistical tendency was found in transfer. The analysis did not report any significant differences for appeals, analytic strategies, and holistic

The Use of Communication Strategies in L3 English CLIL Learners

185

strategies. Even if CLIL II students performed significantly better in the general proficiency test, and consequently, these students were expected to show less instances of previously known language-based communication strategies and a greater use of TL-based strategies, the results obtained do not support this claim as both groups seem to behave similarly in this respect. In addition, taking into account the total number of words produced for each of the groups and the ratio of communication strategies per total number of words, it is possible to claim that the older students were not trying to complete the task with longer or more detailed narrations than their younger counterparts. In fact, younger students, even if their proficiency was lower, produced longer narrations than the CLIL II group. As a consequence, the CLIL II group, due to their higher proficiency, should have resorted to communication strategies less frequently than the CLIL I group. This finding could be accounted for by the fact that the gap in proficiency between the groups is not so big to yield statistically significant differences. If the difference in proficiency level between the groups tested was bigger, the comparison would probably return statistically significant differences. In this regard, this study aligns with recent investigations on self-reported use of communication strategies by CLIL learners (Martínez-Adrián et al., 2019). We may also speculate that the findings in this study could be due to the fact that the difference in the number of hours of exposure between the two groups is scarce, and thus, statistically significant differences do not emerge. Probably, if the difference in the amount of hours of exposure had been bigger, more statistically significant differences could have been found. However, it is true that even if the differences do not reach statistical significance, except for transfer, there seems to be a trend in which the use of previously known language-based strategies is being minimized. These results support previous studies carried out in CLIL settings (Celaya & Ruiz de Zarobe, 2010; Lázaro Ibarrola & García Mayo, 2012; Martínez-Adrián & Gutiérrez-Mangado, 2015) that have considered proficiency as a predictor for transfer. On the other hand, the use of TL-based strategies, usually attributed to more advanced learners in the literature (Bialystok, 1983; Jourdain, 2000; Wannaruk, 2003), does also seem to decrease from CLIL I to CLIL II. In addition, the present study has also reported very interesting data as regards the source language used in appeals for assistance and transfer. The examination of the data showed the inexistence of L2 Basque influence, except for one case in transfer. This result could be due to the greater typological distance between Basque and English when compared to the distance between Spanish and English, and also because of the context in which participants from this study live, where the minority status of Basque would make Spanish the default language (Cenoz, 2009; MartínezAdrián & Gutiérrez-Mangado, 2015). As for the second research question (Which is the predominant communication strategy in both groups?), holistic strategies were found to be the most common category, with statistically significant differences between this category and transfer in both groups. In the case of the CLIL II group, holistic strategies even showed significant differences with respect to interactional strategies, and analytic strategies also surpassed the use of transfer lapses. Even if these learners are still beginners,

186

I. Arratibel-Irazusta and M. Martínez-Adrián

they are found to use TL-based strategies, more typical of advanced learners4 . These results align with self-reported opinions from primary school learners (MartínezAdrián et al., 2019), in which the effect of proficiency is suggested to be overruled by the effect of CLIL. As in the present study, the use of TL-based strategies ranked higher than previous linguistic experience-based strategies. The input learners receive in these meaning-oriented educational contexts may lead learners to manipulate more and to exploit the TL. Nevertheless, even if this is the tendency in the case of the categories under investigation in the current study and in Martínez-Adrián et al. (2019), this is not the case in other studies that have analyzed the use of discourse markers (Arratibel-Irazusta & Martínez-Adrián, 2018; Lázaro Ibarrola, 2016; Lázaro Ibarrola & García Mayo, 2012). Although the more natural input received in CLIL may lead to a lower use of previously known languages in appeals for assistance as well as categories such as borrowings and foreignizings, this is not accompanied by an increase in English pragmatic discourse markers. It is also worth mentioning that despite the predominance of holistic strategies, appeals for assistance are still common among learners, particularly in the case of CLIL I learners, a finding also reported in previous CLIL research (i.e., Martínez-Adrián et al., 2019). Finally, with reference to the third research question (Are there any correlations between general proficiency, receptive vocabulary, and the use of communication strategies (i.e., previously known language-based and TL-based strategies)?), the analysis yielded a statistically significant positive correlation between general proficiency and knowledge of receptive vocabulary in both groups which indicates that as students become more proficient in the TL, they also acquire new vocabulary. Statistically significant negative correlations were found between general proficiency and the use of previous linguistic experience as a communication strategy, which suggests that as the level of proficiency increases, the use of previously known languagebased strategies decreases, a result in line with previous investigations in this respect (i.e., Lázaro Ibarrola & García Mayo, 2012; Martínez-Adrián & Gutiérrez-Mangado, 2015). In addition, a significant negative correlation was observed between knowledge of receptive vocabulary and the use of previously known language-based strategies in the younger group, which implies that level of vocabulary also concurs with the decrease in the use of previously known language-based strategies. The inexistence of a significant correlation in the older group could be explained in the light of the high scores already obtained by this group in the 2000 band. We may think that in a higher band, significant correlations could emerge. As regards the use of TL-based strategies, no significant correlations were found between the use of these strategies, the level of proficiency (Quick Placement Test) and receptive vocabulary (Vocabulary Level Test) in either group. This seems to indicate that the development of TL-based strategies does not concur with the level of proficiency or receptive vocabulary in these young CLIL groups. This finding could be related to the overriding effect of CLIL as we already claimed in response to the second research question. Even if they are beginner learners, they use TL-based strategies in similar or higher proportions 4 See

the studies by Poulisse et al. (1990) in which holistic strategies were used to a higher extent by more advanced learners and transfer by less proficient learners in storytelling and interviews.

The Use of Communication Strategies in L3 English CLIL Learners

187

than previously known language-based strategies. In addition, the examination of the data did not report significant correlations between the use of previous linguistic experience-based strategies and TL-based ones. Thus, the lower use of previously known languages does not automatically lead to an increase in TL-based strategies. This means that previously known language-based strategies still coexist with the use of TL-based strategies. This is consistent with other investigations carried out in CLIL contexts in which a lower use of the L1 on the part of CLIL learners did not lead to an increase of the TL in all the contexts analyzed (Arratibel-Irazusta & Martínez-Adrián, 2018; Lázaro Ibarrola & García Mayo, 2012; Pladevall Ballester & Vraciu, 2017). Previously known languages could be regarded as cognitive tools and to have a regulatory function even as L2 proficiency increases in CLIL settings (Pladevall Ballester & Vraciu, 2017).

7 Conclusion This study aimed to investigate the use of previous linguistic knowledge (L1 Spanish/L2 Basque) as a communication strategy (appeals for assistance and transfer) together with TL-based strategies in an oral narration task in two different age/proficiency CLIL groups of L3 English learners. Proficiency did not have a big impact on the use of these strategies as no differences were found between the groups analyzed. As regards the distribution of strategies, holistic strategies were the most used, which were explained by the overriding effect of CLIL as reported in previously conducted research with young learners (MartínezAdrián et al., 2019).5 With respect to correlation analyses, although previously known language-based strategies do seem to align with the level of proficiency and receptive vocabulary, this was not the case with the use of TL-based strategies, which could be explained by the overriding effect of CLIL. Additionally, despite the use of TL-based strategies by these beginner learners, they do not imply a lower use of previously known languages as a communication strategy. The use of prior linguistic knowledge is still relevant in these CLIL learners, who seem to use it to scaffold L2 production. In future studies, it would be interesting to further analyze the use of communication strategies (both previously known language-based strategies and TL-based strategies) longitudinally so as to shed more light on how the use of these strategies evolves over time. Likewise, studies that compare CLIL to non-CLIL learners would allow researchers to have a more thorough understanding of the effects of exposure to meaningful input in CLIL lessons. In addition, it would be advisable to have information about the teacher’s use of the L1 in class and the amount of communicative activities carried out in the classroom. Similarly, the study of the effect of individual differences such as gender on the use of communication strategies is timely as gender 5 As

suggested by an anonymous reviewer, in a follow-up study, a comparison with a non-CLIL group would be advisable in order to strengthen this claim.

188

I. Arratibel-Irazusta and M. Martínez-Adrián

differences tend to diminish in CLIL settings in other areas such as vocabulary (Heras & Lasagabaster, 2015). Finally, mixed methods research would undoubtedly lead to draw more definite conclusions as regards the use of communication strategies in CLIL contexts.

8 Pedagogical Implications In the light of the results obtained, some pedagogical implications are worth mentioning. Although these learners already use holistic strategies presumably thanks to the effect of CLIL, analytic strategies were not so widespread. As the level of proficiency does not seem to concord with the development of TL-based strategies, training in the use of analytic strategies would report benefits in CLIL classes. In particular, it would be wise to foster this type of strategies during unplanned discourse, when learners do have a greater need to resort to them. Explicit instruction can help in the promotion of an ample variety of strategies. It is not a matter of boosting more effective strategies, but of combining different strategies that may lead to more optimal communication. According to Faucette (2001), active and effective language learners usually make use of appropriate strategies to learn, whereas ineffective learners are worse when choosing and using the appropriate strategies. By explicitly teaching these strategies, learners can become aware of how to use such strategies appropriately to overcome communication difficulties and to result in a more effective communication (Oxford, 1990). In this vein, future research should examine the teachability of communication strategies in CLIL settings (Martínez-Adrián et al., 2019) with the final aim of obtaining a greater assimilation of language and content. In addition, the use of previously known languages is still common among these learners, in particular, the use of appeals for assistance, which demonstrates the ability of the learners to interact (Lázaro Ibarrola, 2016). Taking into account that the use of previously known languages as a communication strategy used for regulatory functions is still common not only in this study but also in other investigations that have analyzed self-reported opinions and oral data (i.e., Lázaro Ibarrola & García Mayo, 2012; Martínez-Adrián, in press; Martínez-Adrián & Gutiérrez-Mangado, 2015; Martínez-Adrián et al., 2019), as well as in studies that have examined observation data from CLIL classes (Gené Gil et al., 2012), the use of previous language experience should be reconsidered in CLIL contexts as a potential resource for educational purposes (Lin, 2015; Swain & Lapkin, 2013), and even when we observe that the emergence of TL-based strategies is also taking place. As Cook (2001) argued, if we consider L2 learning classrooms as bilingual settings, the use of previously known languages should not be forbidden, since this simply makes the language environment in class unauthentic. Thus, they should not be seen as a threat, but as an asset. Despite so, research should be aimed at investigating how both the previously known languages and L2 can work together to develop students’ bilingualism and biliteracy (Tedick & Weseley, 2015). In this sense, informed protocols on how and

The Use of Communication Strategies in L3 English CLIL Learners

189

when to use these languages in CLIL lessons are clearly advocated. Otherwise, we run the risk of not contributing to the correct development of the L2 (Pavón & Ramos, 2019). Acknowledgements The authors would like to acknowledge the grants awarded by the Spanish Ministry of Economy and Competitiveness (FFI2012-32212) and (FFI2016-74950-P) (AEI/FEDER/UE), the University of the Basque Country UPV/EHU (UFI 11/06) and the Basque Government (IT904-16).

Explicit Plurilingualism in Co-taught CLIL Instruction: Rethinking L1 Use Iris Milán-Maillo and Elisabet Pladevall-Ballester

Abstract Language studies framed within sociocultural perspectives have claimed that the use of the first language (L1) in foreign language (FL) instruction settings can facilitate FL learning. This study inquired into the functions of L1 use in the oral production of Catalan/Spanish bilingual primary school learners (N= 20) and teachers (N= 2) of co-taught Content and Language Integrated Learning (CLIL) science sessions. Eight complete lessons that were taught by means of both the L1 (Spanish/Catalan) and the FL (English) were audio recorded and transcribed. The identified functions were quantified and classified into different interactional strategies so as to know which ones were mostly used by which participants. The direction of interaction between addressees and addressers was also analyzed to investigate if the lessons were teacher or student centred. The main findings indicated that this specific co-teaching model did not fulfill the objectives of CLIL approaches but L1 use was shown to be a beneficial tool serving the purpose of coping with CLIL linguistic and cognitive demands. Finally, pedagogical implications are discussed with regard to CLIL and collaborative teaching methodology. Keywords L1 use · Foreign language learning · CLIL · Co-teaching · Sociocultural theory · Interactional strategies

I. Milán-Maillo (B) Department de Tecnologies de La Informació I Les Comunicacions, Universitat Pompeu Fabra, Roc Boronat 138, 08018 Barcelona, Spain e-mail: [email protected] E. Pladevall-Ballester Departament de Filologia Anglesa I de Germanística, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_10

191

192

I. Milán-Maillo and E. Pladevall-Ballester

1 Introduction The use of the first language (L1) within foreign language instruction contexts is often discouraged. Although the L1 is a resource that both teachers and students utilise, it is not an overt and evident practice, but a rather cautious one mainly due to traditional models of language learning neglecting its use. The predominant view encompassing foreign language learning has developed a series of recommendations and policies inviting to teach only by means of the language that has to be learnt, arguing that the presence of other languages during instruction can interfere with the foreign language (FL) learning process (Jones, 2010; Littlewood & Yu, 2011; Pavón & Ramos, 2018). The aim behind monolingual approaches to language teaching is to provide students with large amounts of comprehensible input trying to simulate linguistic immersion and maximise FL use. Studies framed within sociocultural theories (Alegría de la Colina & García Mayo, 2009; Antón & DiCamilla, 1998; Jones, 2010; Swain & Lapkin, 2013; Vanderheijden, 2010) have shown that the prohibition of L1 use might have a negative effect on students’ cognitive learning. In addition, the prohibition of L1 use might entail loss of confidence, affect students’ achievement and trouble the establishment of interpersonal relationships (Jones, 2010). In the same vein, these studies attribute an important role to L1 use when promoting FL learning (Méndez & Pavón, 2012; Pavón & Ramos, 2018) since it is one of the major tools that may encourage and scaffold FL communication. From a sociocultural perspective, language is considered a mediating cognitive tool that regulates thinking and fosters the building of knowledge (Lantolf & Thorne, 2007). This view of language is tightly connected with the objectives of the Content and Language Integrated Learning (CLIL) approach, where language use ideally mediates cognitive processes and aids its learning. Instruction settings that are created thanks to the application of this approach are perfectly aligned with sociocultural claims. Hence, the presence of the L1 in CLIL contexts could play a role in helping FL learners to cope with the cognitive effort invested in their learning process. To date, there is a small number of studies examining the functions of the L1 in co-taught CLIL contexts (Méndez & Pavón, 2012; Pavón & Ramos, 2018). Understanding the different uses of the L1 that are indeed present in the input in FL settings is necessary to discover if FL learners can actually benefit from plurilingual input, and, as an ultimate goal, if they are able to develop a series of strategies through their L1 that could help in facilitating FL development. Therefore, this study analyses the different functions fulfilled by L1 use both by teachers and students in a plurilingual co-taught CLIL context in order to understand the impact of the L1 during FL learning. The present study is organised into six sections. Section 1 presents sociocultural perspectives on L1 use, CLIL approaches to FL teaching, L1 use studies in non-CLIL and CLIL contexts, and the use of L1 interactional strategies. Research questions are addressed in Sect. 2 and the methodology of the study is described in Sect. 3. The results are presented in Sect. 4 and discussed in Sect. 5. The main conclusions and the pedagogical implications of the results are presented in Sect. 6.

Explicit Plurilingualism in Co-taught CLIL …

193

1.1 Sociocultural Perspectives on L1 Use Within the framework of the sociocultural theory, studies consider language as a primary psychological means of mediation and regulation of both mental processes and social interaction (Lantolf & Thorne, 2007; Vygotsky, 1978). Language regulates and mediates the relationships that are established between the individual (biological, interpsychological and behavioural) and the material world (cultural activities, perceptions and intrapsychological relations). Consequently, language is no longer perceived as a simple communication tool, but rather as a key element that is linked to psychological processes that derive its “functions from social activities” (Swain & Lapkin, 2000, p. 254). Language plays the role of being “a resource for participation” (Zuengler & Miller, 2006, p. 37) and a useful device for creating collaborative activities. Thus, language learning takes place in settings where language use is involved, so language learning is both a cognitive and a social process (Swain, 2000). Following these assumptions, one could argue that the L1 can act as a useful cognitive tool in mediating FL learning and responding to specific communicative needs when coping with linguistic and cognitive demands. In other words, the L1 can be useful in mediating the expression and the construction of both knowledge and content in FL instructional settings. This construction of knowledge would originate from the dialogue among the collective on the social plane. It is especially in cognitively demanding activities when learners co-construct meaning, spontaneously externalise their thoughts, and interact with their peers by means of the L1, since it is the language that students share at the same level of competence. This kind of interaction that is framed within Vygotsky’s dialogic model (Swain & Lapkin, 2013)1 is labelled as private speech, and it allows learners to overcome the linguistic difficulties they might encounter when doing a task.2 L1 private speech serves the purpose of developing three important functions when learning a FL: metacognitive and cognitive functions (monitoring the task and its goal), social functions (sharing understanding of the task) and affective functions (externalising thoughts and feelings). In sharing competence and serving these functions, the L1 allows learners to discuss and perform the tasks at higher levels. Sociocultural perspectives agree on the fact that the L1 provides the learner with cognitive support when students are not able to make progress in a given task by using the FL (Castellotti, 2001). The use of the L1 allows FL learners to “establish a shared understanding of the task, to set task goals, and to comment on their L2 use (metatalk)” (Alegría de la Colina & García Mayo, 2009, p. 327). Together with these functions, the L1 also helps the students to make progress within their zone of proximal development (from now on ZPD). As described in Vygotsky’s (1978, p. 86) study, the ZPD is “the 1 This

refers to collaborative dialogue and the notion that individuals derive knowledge from social interaction. It is by the interaction of two or more individuals that language learning and knowledge development can (co-)occur while collaborative dialogue is taking place. 2 This does not mean that the FL or a mixture of both the L1 and the FL cannot be used in private speech. It is likely that learners use a mixture of both languages during their learning process, especially as the proficiency level increases.

194

I. Milán-Maillo and E. Pladevall-Ballester

distance between the actual developmental level as determined by independent problem solving and the level of potential development as determined through problem solving under adult guidance or in collaboration with more capable peers”. This concept reinforces the idea of deriving knowledge and scaffolding from communicative tasks, both enhancing co-construction of meaning and promoting natural language use, social interaction and collaborative dialogue. Thus, CLIL contexts alongside their communicative purposes and the social interaction that is derived from them perfectly embody and prompt learners’ development within their ZPD. Following these assumptions, the presence of the L1 would act as a facilitator element when coping with cognitive, psychological, linguistic and socio-affective difficulties in FL settings.

1.2 CLIL and Co-teaching Similar to the conception of language within sociocultural frameworks, CLIL settings were also designed to mediate (language) learning through language use and to reinforce learners’ cognitive capacities when processing language. The essence of this approach is to teach certain subjects of the curriculum through the use of a foreign language, which means that both content and target language are assumed to be learnt at the same time (Coyle, Hood, & Marsh, 2010; Dalton-Puffer, 2011; Lyster, 2007). This model satisfies the emphasis given by the European Commission on trying to soften the lines of the existing communication gap among the member states of the European Union in the name of inclusiveness, diversity and participation. In relation to this political agenda, new educational policies have started to be applied in order to fulfil the demands posed by the European Commission’s multinational research project of creating a bilingual and multilingual Europe (Commission of the European Communities, 2006). Nevertheless, the emergence of CLIL can also be seen as a response and a possible solution for the previous foreign language learning models, since its aim is to provide relevant input beyond the layer of pure language instruction and to overcome learners’ low levels of proficiency (Muñoz, 2007). In embracing the dual focus of integrating content and language rather than only focusing on language itself, CLIL instruction has positively attributed the same importance to content and language, both being learnt hand in hand (Lyster, 2007). Learners, in being engaged with the content matter, presumably create more positive attitudes towards language use and learning, which might help them to develop linguistic competence and effective communication (Lasagabaster, 2008). When it comes to CLIL implementation, a wide variety of teaching modalities emerge, from those in which the L1 and the FL are used to different extents (fostering plurilingualism) to totally monolingual foreign language CLIL classes. Plurilingual CLIL practices, which are crucial for this study, consist of two or more languages coexisting in CLIL lessons. The L1 is present together with the FL when teaching and discussing the content of the subject matter. In these kinds of contexts, the L1

Explicit Plurilingualism in Co-taught CLIL …

195

is available for the students in classroom settings and constitutes an important tool favouring learning development (Méndez & Pavón, 2012; Pavón & Ramos, 2018). The co-existence of the L1 and the FL within the same instructional context might give rise to a situation in which both languages do not need to be used separately depending on the situation, but rather can be present in the same one, creating a plurilingual environment. This context might be handled by collaborative teaching, that is to say, by two different teachers giving instruction at the same time (Méndez & Pavón, 2012). Collaborative teaching can also be applied in different ways. For instance, one teacher can be in charge of content, whereas the other one can be in charge of language instruction. Another possibility within collaborative teaching is to assign to each teacher a language of instruction. One of them would be delivering the lesson by means of the L1, whereas the other one would be doing so by means of the FL. This teaching modality would be able to fulfil, at least theoretically, the gaps highlighted both by sociocultural perspectives and research on CLIL contexts. Concerning sociocultural theories, co-taught CLIL instruction would be able to cope with learners’ linguistic and cognitive demands, for the L1 would be available and embodying the role of knowledge mediation. In line with this, but more strictly related to CLIL contexts, collaborative teaching reinforces the idea of integration. Besides, it is clear that a “development of both language and thinking skills is needed for content and language to be acquired in communication” (Basterrechea & García Mayo, 2013; cited in Devos, 2016, p. 14), for CLIL lessons require higher levels of both linguistic and cognitive processing due to its cooperative nature. Hence, by applying this co-teaching model, teachers’ provision of language and thinking skills is reinforced by the presence of the L1. With respect to students, the dialogic activity that might be prompted by CLIL pedagogy (e.g., cooperative learning activities, group-work) may be more easily tackled because of the availability of the L1. The use of the L1 in co-taught CLIL lessons might allow students to link knowledge and communication (Devos, 2016). Therefore, the L1 would act as a strategy with the objective of fulfilling both language and content gaps. Referring to instruction and direction of interaction, two different processes might take place. On the one hand, the direction of interaction would be from individual to individual/collective (students). In this case, content and language would be integrated and embodied by the teacher (as an individual) and mediated through language towards the student(s). On the other hand, both teachers as a collective would integrate content and language (and also between them) and thus, the interaction would be from the collective to the individual/collective of students. The fact that each of them mediates integration by means of a different language system would not disturb or prevent interaction from taking place, but it would probably prompt different inputs on the teachers’ part and different outcomes derived from the learners’ learning process. Studies on L1 use in co-taught plurilingual CLIL settings are scarce but those framed in CLIL and English as a Foreign Language (EFL) contexts are more and more frequent. Some of the most relevant ones, particularly in relation to our local context, will be reviewed in the next section.

196

I. Milán-Maillo and E. Pladevall-Ballester

1.3 Previous Research on L1 Use in CLIL and Non-CLIL Contexts A number of studies both in CLIL and non-CLIL contexts have examined L1 use as a mediating tool in FL learning and use. In a non-CLIL context, Antón and DiCamilla (1998) examined the nature of learners’ collaborative talk and strategies when dealing with writing tasks. Learners worked effectively towards problem resolution and constructed a shared perspective on the task through their L1, “a powerful tool of semiotic mediation between learners (at the interpsychological level) and within individuals (at the intrapsychological level)” (Antón & DiCamilla, 1998, p. 234). Results showed that the collaborative dialogue carried in the learners’ L1 fostered comprehensible input in the FL and prompted acquisition. Along these lines, the study seemed to validate the argument that the L1 has to be established as an important psycholinguistic and cognitive tool that mediates and enhances FL social interaction and learning. In similar ways, other studies (García Mayo & Lázaro-Ibarrola, 2015; Lázaro-Ibarrola & Azpilicueta Martínez, 2015) have shown that the L1 in EFL contexts has been a powerful tool for resolving communication breakdowns. Authors such as Vanderheijden (2010) explored the properties of L1 use in non-CLIL classroom settings, and concluded that the native language was a facilitating tool for both language learning and noticing. The author discussed predictable learners’ L1 use in three different situations (i.e., information gap, narration jigsaw and cloze texts tasks) and the different functions that the L1 might have as well. In the study, learners used their L1 for controlling and managing the task. Furthermore, the L1 helped learners to process the information that was given in the second language (L2). Similarly, Alegría de la Colina and García Mayo (2009) reported that students resorted to their L1 to self-correct, ask for help and manage or discuss the tasks they had to complete in the L2. Basically, L1 use was a resource that provided cognitive support and served the purpose of developing learners’ understanding and facilitating L2 input processing. Tognini and Oliver (2012) also highlighted the importance of L1 functions in foreign language learning settings. In the study, both researchers discovered that L1 use had positive social and cognitive functions deriving from peer interaction. The study also proposed that L1 use fulfilled different functions that allowed learners to move the task along, focus attention on the task, create interpersonal interaction and develop learners’ identity. Nonetheless, the study also pointed out the need to balance L1 and L2 use to benefit learning. The situations in which the L2 was used were very much restricted due to both the presence of the L1 and the L2 low proficiency on the learners’ part. With regard to CLIL contexts, Lázaro-Ibarrola and García Mayo (2012) evaluated the role carried out by the L1 in CLIL settings in order to characterise CLIL discourse features (e.g., discourse markers and repair sequences) and morphosyntactic development when learners narrated a story. The study suggests that L1 use was crucial when learners asked for help during narrative production. In addition, children were allowed to make use of the L1 whenever they recognised a gap between what they wanted to say and what they were able to express (ZPD). By using the L1, learn-

Explicit Plurilingualism in Co-taught CLIL …

197

ers were able to mediate their knowledge through collaborative dialogue. Results suggested that learners were not able to cope with linguistic difficulties in English. Learners did not use English for expressing discourse markers or when paraphrasing sentences/words, a characteristic that supposedly would differentiate CLIL learners from regular EFL learners. Other studies have compared oral production tasks in non-CLIL versus CLIL contexts and have explored if type of instruction has any effects on L1 use. Gené Gil, Juan Garau, and Salazar Noguera (2012) examined L1 and FL functions of both teachers and students in a CLIL setting as compared to an EFL setting. The study was conducted by means of questionnaires, oral interviews and classroom observation. The authors observed that there were more instances of code-switching and L1 use in CLIL instruction with respect to EFL lessons, which might be attributable to the difficulties posed by the content subject. The fact that teachers did not restrict students’ L1 use might have also been an important factor. Findings also highlight that the L1 did not compete with the FL and that is was used as a support and reinforcement tool that benefitted the students’ learning. Pladevall-Ballester and Vraciu (2017) examined L1 use in oral narratives in CLIL and EFL contexts. The study examined the way in which learners utilised their L1 as a resource when coping with the linguistic and cognitive demands of a narrative oral task. Results indicated that CLIL learners made use of L1 interactional strategies, being metatalk and private speech the most recurrent ones. In addition, the L1 was acknowledged to be fulfilling the functions of self-regulation and scaffolding of FL production, although the scaffolding function decreased as proficiency level increased. Nevertheless, the L1 was analysed as a compensatory strategy for FL learners, regardless of the instructional setting. Similarly, García Mayo and Hidalgo (2017) conducted a longitudinal study exploring L1 use in mainstream foreign language lessons and CLIL. They used a two-way communicative task in which the students had to complete a poster with the information provided by their partner, fostering interaction on students’ part. Results showed that students relied minimally on their L1 during the experiment and that students switched to their L1 to facilitate task completion, providing further evidence of the fact that the L1 can act as a scaffolding tool. Martínez-Adrián (2018) compared L1 use in interactional strategies of two CLIL and non-CLIL groups (grades 4 and 6) of primary school children. The interactional strategies under study were appeals for assistance, clarification requests and metacomments. As for non-CLIL learners, the use of the L1 was found to be more frequent in metacomments strategies, whereas 4th grade CLIL learners preferred to use appeals and 6th grade CLIL learners preferred clarification requests. Nonetheless, results suggest that for both groups the L1 was mainly used when organising and monitoring the activity. Martínez-Adrián, Gallardo-del-Puerto, and Basterrechea (2017) inquired into primary school learners’ (grades 5 and 6) self-reported opinions about their use of communication strategies by means of questionnaires. Analyses showed that L2-based strategies were generally favoured in both CLIL groups. However, L1-based strategies were also significantly used in grade 6. As a whole, results suggest that resorting to the L1 did not interfere with their target language

198

I. Milán-Maillo and E. Pladevall-Ballester

learning. Arratibel-Irazusta and Martínez-Adrián (2018) examined L1/L2 (Spanish and Basque, respectively) use in two groups of secondary school CLIL learners in an oral narration task conducted in the L3. Results showed that 2nd and 4th grade students showed similar amount of L1/L2 use in interactional strategies, borrowings, code-switching and discourse markers. Discourse markers was the most used strategy among groups, which shows that the use of the L1 was framed within the unplanned discourse and served the purpose of facilitating the flow of speech. Finally, Pavón and Ramos (2018) investigated L1 strategies used by primary school learners in a co-taught CLIL setting. Results showed that a wide variety of L1 strategies was used with the main objectives of clarifying meaning, supporting general understanding of ideas, concepts and vocabulary. The L1 was also frequently used to initiate and maintain interaction, thus facilitating the correct use of the FL during challenging situations. Previous research has established the need to further explore L1 use as a communicative and interactional strategy in FL learning situations. The types of L1 interactional strategies used in this study are justified and described in section below.

1.4 L1 Interactional Strategies It is generally acknowledged that learning strategies play a crucial role in FL acquisition. In agreement with the assumption that language is a mediating tool, FL learning strategies can be defined as “a learner’s socially mediated plan or action to meet a goal, which is related directly or indirectly to L2 learning” (Oxford & Schramm, 2008, p. 48). In other words, there is an underlying process of mediation from the social to the individual when using learning strategies, which again points towards the importance of learners’ social interaction for processing L2 knowledge in foreign language instruction (and specifically in CLIL contexts because of their communicative purposes). By the same token, this definition also reinforces Vygotsky’s sociocultural and cognitive theory on collaborative dialogue. The dialogic processes that take place when interpersonal relationships are established enable learning processes to be mediated through the use of strategies. These strategies are thought to allow learners to integrate learning better. Importantly, it needs to be pointed out that dialogic views of language applied to instructional settings give rise to the idea that “different educational purposes require different ways of talking in the classroom” (Moate, 2011, p. 22). Hence, different types of strategies will be needed so as to accomplish different objectives (i.e., understand the concepts of the lesson, ask questions about the language system that is to be learnt, and so on). When trying to communicate in a FL, learners resort to a type of strategy referred to as compensatory strategies, which are used by the speakers with the objective of communicating their intended meaning after detecting the difficulties and problems posed by the communicative situation (Hüttner and Rieder-Bünemann 2010). Within compensatory strategies, L1 interactional strategies emerge as “intentional switches into languages other than the target language” (Cenoz, 2003b, p. 3). Thus, the devel-

Explicit Plurilingualism in Co-taught CLIL …

199

opment, effectiveness and internalization of these strategies mainly depend on the social interaction and the collaborative dialogue that is carried by both teachers and learners through their L1. L1 use is then a compensatory strategy that responds to specific linguistic, cognitive, emotional and communicative needs that have to be resolved in order to cope with CLIL demands, both on the learners and the teachers’ part. More elaboration on this topic will be presented in the discussion section. For the purpose of the study, seven types of L1 interactional strategies following and adapting the classifications of strategies by O’Malley and Chamot (1990), Oxford and Schramm (2008), functions by Alegría de la Colina and García Mayo (2009), and talk-types by Moate (2011) are set out. The classification used for the present paper together with examples found in the data of the present study are presented below. Metacognitive strategies are those that relate to clarifications on meaning, content and objectives, cooperative and assertive interaction with peers aimed at increasing knowledge and exploratory talk (i.e., asking questions committed to building understanding), as illustrated in (1): (1) Algú té alguna idea del que pot passar? [teacher asking whole class] Does anyone have any ideas on what might happen?

Metatalk strategies concern comments, questions and feedback about the language system. Metatalk can be considered a mix of focus-on-form and focus-onmeaning approaches, since one of the objectives of this type of strategy is to foster a correct language use and learning regarding language features and meaning: (2) Find és buscar? [student asking the teacher] Find is look for?

Task-related strategies entail instruction of task procedures, clarification of specific task-based instruction, comments on task procedures and the organisation of the task. This kind of strategy is always framed in the here-and-now context of the classroom environment. It tends to be formulaic, repetitive, and it is based on “maintaining the focus and pace of [the] activity” (Moate, 2011, p. 23). (3) Escolteu queden dos minuts! [teacher addressing whole class] Listen, there are two minutes left!

Pedagogical strategies include translation and a type of talk focused on bridging between everyday understanding and expert conceptualization (leading to decisive higher cognitive basis responses). In this type of dialogue, the expert mediates the learning process of the learner by joining the construction of knowledge that flows out from the conversation of both interlocutors. This dialogic space allows learners and experts to elaborate on their explorations until they find a satisfactory answer. Hence, both participants actively co-construct knowledge. (4) What’s a paraula clau, what’s a keyword? [teacher asking students] What’s a keyword, what’s a keyword?

200

I. Milán-Maillo and E. Pladevall-Ballester

Social/affective strategies involve motivation, control over affective factors (either individual or shared among the learners of a specific social group), off-topic and non-assessed talk between peers, fillers, hesitations and comments about classroom procedures that are not directly related to knowledge building, as (5) shows: (5) Qué chungo. [student-student] How tough (colloquial: what a bummer)

Organisational strategies refer to practical management of classroom procedures (clarifications, commands, questions). Similar to task-related strategies, this kind of strategy also tends to be based in the here-and-now of the classroom environment, as shown in (6): (6) Fem dos grups que facin la mateixa? [teacher-teacher] Shall we make two groups that might do the same?

Self-regulation strategies include planning, monitoring progress, reflecting on performance, setting goals, self-evaluating and restructuring the methods. This type of strategy can identify questions in two directions. The first question would be addressed to the self (e.g., ‘what do I know about the topic?’, ‘how did I perform?’, ‘what is the goal?’). Then, the second question (that can be formulated either in the form of a question or a declarative sentence), which would be a second reflection of the topic under exploration, would allow the learner to participate in the learning process in an active way co-constructing the knowledge with other peers and teachers: (7) Abans no he escoltat [student addressing the teacher] I was not listening before.

2 Research Questions Given the scarcity of research on L1 use in co-taught CLIL instruction settings, this study seeks to contribute to that gap by examining and describing L1 use patterns in the oral production of primary school learners and teachers of co-taught CLIL science sessions. By analysing the nature and the different types and purposes of L1 use, some implications for FL learning will be drawn, for it will be possible to explore if the L1 is a beneficial tool serving the purpose of coping with CLIL linguistic and cognitive demands. Thus, the research questions that this study seeks to answer are the following: 1. What L1 interactional strategies are more frequently used by which participants? 2. What is the direction of participants’ L1 interaction?

Explicit Plurilingualism in Co-taught CLIL …

201

3 Method 3.1 Participants The study was conducted in a local primary Catalan/Spanish bilingual school situated in a newly created neighbourhood in a town near Barcelona. The school started to function during the academic year 2006–2007, but it was not until 2011 that it had a more clearly outlined project together with its new facilities. The institution offers schooling for two educational stages, namely infant school and primary education. The demographic composition of the school is diverse but most of the students are from Spanish/Catalan origins. Many of the students are familiar with the English language, since most of them have attended private English classes and some of them have had babysitters that interact with them in English. The school started to develop a Pla Integrat de Llengües Estrangeres (PILE) in 2012 that promoted a plurilingual environment in different subjects taught in the school. It also prompted the implementation of a co-taught CLIL course on science. The main aim of the plan was twofold. On the one hand, it was designed to offer the possibility of making use of a foreign language in a functional and significant way. On the other hand, the design followed the objective of teaching science by means of a language that is different from the native one. After completing the PILE, the school started to participate in the Grup d’Experimentació per al Plurilingüisme (GEP) project. Despite the development and the methodology of the PILE and the GEP were different, the tenets underlying both programmes were very much the same. Since 2012, the school has developed and encouraged plurilingual environments in instructional settings. The participants of the study were 20 6th grade students, aged 11–12, and two female teachers co-teaching CLIL science lessons. The students received between 40–45 min of CLIL science instruction per week in addition to two EFL hours per week. Both teachers taught the lessons simultaneously, that is, they coexisted in the same space and at the same time. However, both teachers had different language predefined roles, which fostered a plurilingual context. One of the teachers delivered the lessons in Catalan, whereas the other one taught the lessons in English. There were no restrictions concerning the amount of L1 use. The teacher who mediated learning by means of the L1 was the teacher who normally taught regular science lessons at that school. The teacher who delivered the lessons through English was normally in charge of teaching regular EFL lessons. She was supposed to lead the science class (i.e., she was supposed to be the “main teacher”), had received previous training on CLIL, and held a C1 in English language. Both teachers had the same training background and were equally capable of delivering science lessons. However, given their teaching trajectory inside the school, it is clear that the teacher who taught regular science lessons was more likely to contribute to the content part than the teacher who taught regular EFL lessons, and the latter could contribute more to

202

I. Milán-Maillo and E. Pladevall-Ballester

the language part.3 Although none of the teachers had received previous training on co-teaching, both of them complemented each other. As the language teacher was in charge of the classroom, the content teacher intervened either to contribute with specific or highly elaborated content explanations using the L1 or to translate some of the explanations provided by the language teacher. On these occasions, the language teacher took advantage of the situation to provide FL linguistic translations or annotations. Both learners and teachers were informants of the study. Students were always involved in group-work activities and they worked in the same groups in each of the lessons. The main objective of the CLIL science lessons was to develop a scientific experiment linked to the subject curriculum. Every group of students had to design an experiment related to a given topic (e.g., the effect of ultraviolet rays on different materials). First, they had to write a short text introducing the topic (e.g., what ultraviolet rays are and the impact they can have), then they had to design an experiment together with their hypotheses and finally they had to test their hypotheses by carrying out their experiment.

3.2 Procedure The study was based on the oral production data collected during classroom observation. To collect data for this study, eight complete lessons were audio recorded, resulting in 330.19 min of recorded data. Narrative summaries derived from field notes were produced for each lesson. Lessons were also transcribed and coded.

3.3 Coding and Data Analysis The content of each transcription was further classified in line with the research questions requirements. The data were classified according to L1 interactional strategies for quantitative analysis and qualitative description of lesson extracts. L1 interactional strategies were divided into seven categories in line with sociocultural perspectives depending on the function they were fulfilling, namely, metacognitive, metatalk, task-related, pedagogical, social/affective, organisational, and self-regulation. Direction of interaction was also taken into account within this classification. The types of interaction that were considered in this study were the following: content teacher-whole class (CT-WC), language teacher-whole

3 When

referencing their activity and interaction in the classroom setting, both teachers will be differentiated following this assumption, i.e., “content teacher” (abbrev. CT) will be the label for the one who normally teaches regular science lessons and speaks in Catalan, and “language teacher” (abbrev. LT) will be the label for the one who teaches regular EFL lessons and speaks in English.

Explicit Plurilingualism in Co-taught CLIL …

203

class (LT-WC), teacher-teacher (T-T), content teacher-student/s (CT-Ss), language teacher-student/s (LT-Ss), student-student (Ss-Ss) and student-teacher (Ss-T).

4 Results As regards the first research question, all types of L1 interactional strategies were observed in the data, although metacognitive and social/affective strategies are the ones that prevailed in the majority of the participants that used the L1. The seven types of strategies that were investigated (i.e., metacognitive, metatalk, task-related, pedagogical, social/affective, organisational and self-regulation functions) are presented in Table 1 taking into account the direction of interaction when using the strategies. Examples of strategies-framed within the context of group work activity explained in Sect. 3.1 are provided above. As Table 1 shows, the four strategies that are mostly used in CT-WC interaction are the metacognitive (8), social/affective (9) and task-related (10) ones followed by the organisational one (11). The strategies that were least used by this participant were metatalk (12) and self-regulation (13): (8) com que deuen faltar, faltaven connectors, per això a l’hora de dir-ho no s’entenia gaire bé, semblava que (ell) digués, veiem si obrim els ulls, si no els obrim no veiem because some connectors were missing, some connectors were missing, that’s why it was not so clear, it looked as if he said, we can see if we open our eyes, if we don’t open them we can’t see (9) us ajudarem la (language teacher’s name) i jo the language teacher and myself will help you out (10) cada taula pensarà el seu disseny i la seva història each table will think of their own design and their story (11) mirarem de # de… # dissenyar un experiment de manera que ho pugueu comprovar això we’ll try to design an experiment so that you can test this (12) com li diem quan la llum rebota? How do we say it when light bounces off? (13) jo ja no sé com explicar-ho… I don’t really know how to explain this…

In the case of the language teacher, the two strategies that are mostly used when interacting with the whole class are the pedagogical (14) and the organisational ones (15). The strategies that were less prominent were metatalk (16) and metacognitive (117). Interestingly, no instances of task-related and self-regulation strategies are found:

11

278

176

325

Ss-Ss

Ss-CT

Ss-LT

32

T-T

111

2

LT-WC

LT-Ss

204

CT-WC

CT-Ss

Metacognitive

Strategies

57

0

35

9

5

10

1

3

Metalolk

59

36

11

0

62

4

0

117

Task-related

Table 1 Total number of interactional strategies per participants

83

2

71

64

53

14

21

88

Pedagogical

303

100

231

14

89

76

4

153

Social/affective

47

7

46

0

32

20

8

111

Organisational

74

12

44

0

1

1

0

11

Self-regulation

204 I. Milán-Maillo and E. Pladevall-Ballester

Explicit Plurilingualism in Co-taught CLIL …

205

(14) incoming light, yes, que ve # it could be in catalan # la llum que VE, la llum que ENTRA incoming light, yes, which is coming, in Catalan it could be the light that’s coming, the light that’s coming in. (15) guardem i continuarem demà let’s pack and we’ll go on tomorrow (16) how do you say rebota? how do you say bounce off? (17) because what I did # is translating what you said # but she is right # it’s not # el que JO penso because what I did is translating what you said but she is right, it’s not what I think.

Regarding teacher-student(s) interaction, the strategies that were most widely used on the content teacher’s part were metacognitive (18) and social/affective (19) strategies. The less widely used strategies were self-regulation (20) and metatalk (21). As for the language teacher, pedagogical (22) and social/affective (23) were the most widely used strategies. Interestingly, no instances of organisational strategies, task-related and self-regulation strategies are found: (18) CT-Ss: perquè una lupa pot cremar objectes, no? because a magnifying glass might burn objects right? (19) CT-Ss: i llavors m… dius, bueno and then you mm…say, oh well. (20) CT-Ss: clar que té sentit of course it makes sense (21) CT-Ss: no heu posat off, heu posat on you didn’t write off, you wrote on (22) LT-Ss: but do you say lupa in Catalan? o you say… llanterna? but do you say magnifiyng glass in Catalan? or you say…torch? (23) LT-Ss: look look! # molt bé! look look! Well done!

As for teacher-teacher interaction, it is mostly managed in terms of social/affective (24) and metacognitive strategies (25). This indicates that they normally addressed each other for off-topic talk and for cooperative interaction (i.e., define and talk about the content of the task/lesson). Scarce instances of task-related (26) and selfregulation (27) strategies were found in the data. Thus, issues about task procedures and instructions were not discussed between them: (24) CT-LT: no, oi? estic una mica empanada… that’s not it right? I’m a bit absent (colloquial: empty-headed) (25) CT-LT: és que es salten la retina the thing is that they skip the retina

206

I. Milán-Maillo and E. Pladevall-Ballester

(26) CT-LT: això és posterior this comes after that LT-CT: ok after they finish # ok CT: no és la feina que… isn’t that the task that…? LT: I am just asking # ok that’s fine (27) CT-LT: alumini és steel? és que no ho sé…4 steel is steel? The thing is I don’t really know…

For student-student interaction, metacognitive (28), social/affective (29) and pedagogical strategies (30) were the most widespread strategies. The strategy that was used the least was task-related (31). Although the difference between metacognitive and social/affective strategies is not very large (see Table 1) it is interesting that students used to employ their L1 primarily to increase their knowledge and cooperate with their peers during the tasks. In line with that, students also used their L1 to bridge the gap between everyday understanding and expert conceptualisation among themselves. However, they did not usually use their L1 to discuss about task procedures: (28) Ss-Ss: home, evidentment que la travessa, perquè mira… come on, it’s clear that it goes through, look Ss-Ss: sí, reflecta yes, it reflects (29) Ss-Ss: m’he equivocat oi, llegint? que he llegit malament tio I made a mistake while reading didn’t I? I did it wrong man (30) Ss-Ss: ultraviolet rays may burn some materials # que…els rajos ultraviolet # calenten # alguns materials ultraviolet rays may burn some materials, that ultraviolet rays heat some materials Ss-Ss: que cremen they burn them (31) Ss-Ss: ja però és que… # així directament? yes but…just like that?

When it comes to student(s)-teacher interaction, students mainly used metacognitive (32) and social/affective (33) strategies when addressing both the content and the language teacher. Nevertheless, the strategy following these two is different depending on the addressed teacher. Whenever students interact with the content teacher, L1 4 The part of the sentence that is emphasised in italics has been analysed as self-regulation. The first

part of the sentence is included in order to understand the situation better, otherwise “és que no ho sé” could be interpreted as part of a metacognitive strategy. It is argued here that this sentence belongs to self-regulation, because the content teacher intervention is not fully focused on clarifying the meaning of the word “steel”. Rather, it is more focused on self-reflecting on her lack of knowledge in relation to a given word.

Explicit Plurilingualism in Co-taught CLIL …

207

discourse focuses on task-related (34) issues, whereas the interaction with the language teacher is more pedagogically centred. Pedagogical (35) and organisational (36) strategies were not widely used when addressing the content teacher. It is also worth pointing out that students do not use metatalk strategies when interacting with the content teacher. Metacognitive (37), social/affective (38) and pedagogical strategies (39) were used when addressing the language teacher, and less widely used was the organisational (40) strategy: (32) Ss-CT: i es poden cremar els líquids? and can we burn liquid? (33) Ss-CT: no, és una ferida no, it’s an injury (34) Ss-CT: apuntar-los i fer una exposició sobre l’experiment (you need to) note them down and do an oral presentation about the experiment. (35) CT-Ss: estareu molta estona esperant-vos a veure si travessa? are you going to wait long to see if it goes through? Ss-CT: no! Ss-CT: o sigui, si travessa és directament! so, if it goes through it will do so right away! (36) Ss-CT: però això ho farem o no? o és com… but are we going to do that or not? Or is it…? (37) Ss-LT: el que rebota, el que refracteix The one that bounces off, the one that refracts (38) Ss-LT: que no paren de parlar de l’amic invisible they are always speaking about secret Santa… (39) Ss-LT: find és buscar? find is look for? (40) Ss-LT: estem, és, estem parlant de quin dia podem quedar per… we are, is, we are talking about what day we could meet to…

As regards the second research question, total percentages regarding the direction of L1 interaction are shown in Fig. 1. Percentages were calculated out of the counts presented for interactional strategies. The majority of L1 interaction occurs when students address teachers, and specifically, the language teacher.5 Teacher-whole class is the second classification that 5 Students used to interact with the language teacher much more than with the content teacher. Then,

as students addressed teachers in their L1 almost by default, the percentage of students-language teacher interaction is much higher than with the content teacher. Hence, it is not the case that the students made an effort for addressing the content teacher in the FL (which might have been a fair interpretation derived from the data presented in Fig. 1).

208

I. Milán-Maillo and E. Pladevall-Ballester

Fig. 1 Total percentage of L1 interaction concerning direction

predominates. As expected, the content teacher leads this category, since the language teacher mainly communicates with students by using the FL. The category that follows is interaction between students, representing less than a quarter of the total interaction. These results will be discussed in the next section.

5 Discussion This study has examined the functions of L1 use in the oral production of CLIL science primary school learners and teachers. Concerning the first research question, the use of L1 interactional strategies is varied among participants and seems to indicate the functions and roles that each participant had in the classroom. The fact that both teachers used different strategies in a mutually exclusive way indicates that their roles were very well defined. This separation of roles is also exemplified in the cases in which teachers used the same strategies when addressing the whole class or a group of students (i.e., the content teacher did not use metatalk and self-regulation in either case). Thus, it seems that teachers had separate roles not only in terms of language, but also in terms of class management. As for teacher-whole class interaction, the content teacher was the one who led the lesson in terms of clarifying the objectives and the contents of the lesson, increasing students’ knowledge of the subject as well as making sure that the established procedures were straightforward. The content teacher did not normally use the L1 to deal

Explicit Plurilingualism in Co-taught CLIL …

209

with language issues or monitor students’ progress, but rather she focused on content delivery and instruction clarification. As she was only concerned with content and not with language, both her role and her figure did not adjust to the objectives of the CLIL approach, since CLIL ideally fosters content and language learning to the same extent. As for the language teacher, pedagogical and organisational strategies were the most prominent ones. The way in which the language teacher used the L1 is in line with the extracts presented in Pavón and Ramos (2018), where teachers avoided using students’ L1, but when using it, they mainly did it for pedagogical purposes. The language teacher addressed students whenever she felt that there was some misunderstanding or language problem going on. When dialogues between her and students occurred, she adopted the role of the expert, mediating the learning process of the students and prompting co-construction of knowledge. In addition, she would translate the concepts that were not understood into the L1 and clarify classroom organisational procedures. Likewise, she also interacted with the whole group in order to develop students’ knowledge at a higher cognitive basis and comment on organisational procedures. The use of these strategies is partly in line with findings from Qian (2009), who reported that teachers used their L1 to mainly develop metacognitive, metatalk and social strategies so as to provide necessary scaffolding and optimal conditions to promote FL learning. In this study, however, none of the teachers widely used metatalk strategies when addressing students. As teachers’ use of the L1 was not focused on reflecting on the language system, the potential of the L1 in helping in the FL learning process was diminished. Task-related and self-regulation strategies were not used at all (both in LT-WC and LT-Ss interaction) and organisational strategies were only used in whole-class interaction situations. It appears that the language teacher did not fulfil the functions of dealing with task-related and organisational issues because it was the content teacher that mainly embodied them. It is worth noticing that the language teacher never used the L1 to monitor the progress of the students or reflect on students’ performance. Consequently, the potentiality of using the L1 for regulating and mediating learning remained ignored. In contrast, the content teacher did use the L1 for this purpose, but only on few occasions. At least, as far as the L1 is concerned, it seems that the language teacher did not play an important role with regard to classroom management and organisational procedures. In relation to teacher-teacher interaction, the fact that teachers barely interacted between them to talk about task management reinforces the division of the roles that were established and indicates that classroom and task management mainly involved and pertained to the content teacher. The division and the collaborative teaching model that has been presented seem not to really adjust to the CLIL approach and principles, for the model does not incorporate real integration. This might be attributable to the fact that the two teachers did not have many opportunities outside the classroom to discuss and model the collaborative situation. Moreover, none of the teachers had received previous training on collaborative teaching and hence they were almost forced to rely on their intuition. These arguments coincide with those of Méndez and Pavón (2012), who observed that teachers participating in collaborative

210

I. Milán-Maillo and E. Pladevall-Ballester

teaching programmes did not have proper guidelines for effective co-teaching, which is detrimental to the successful management of plurilingual instruction. Regarding students, communication among them was mainly based on cooperative interaction, but also on non-assessed talk. This situation poses L1 use into two opposite extremes, since—in broad terms—the L1 either served the purpose of clarifying the meaning or the content of a given task or talking about topics that were not of primary interest regarding the lessons. Thus, on these occasions the L1 functioned merely as a communicative tool through which students could interact with their peers without restrictions. As they could interact by using the L1 as much as they wanted, there seemed to be no need to try and produce FL output to communicate, which had also been observed by in Tognini and Oliver’s (2012) study. However, it was seen that students used their L1 for pedagogical purposes too, which developed learners’ understanding and facilitated FL input processing (Cummins, 2007; Littlewood & Yu, 2011; Martínez-Adrián, 2018; Pavón & Ramos, 2018; Tognini & Oliver, 2012). Whenever students used metacognitive and pedagogical strategies they were building on their knowledge through dialogic activity mediated by their L1. Therefore, the L1 fulfilled an important cognitive and social strategic role, for learners were capable of regulating their learning through strategies during peer interaction. Similar results are shown in Martínez-Adrián (2018), where the greater amount of L1 use of grade 4 CLIL learners occurred during appeals for assistance (equivalent to our metacognitive strategy). However, it is important to highlight that task-related strategies were not widely used among the students of this study, which is not in line with the results in Martínez-Adrián (2018). This suggests that students did not discuss about task procedures, which is rare given the collaborative nature of the classroom setting. This might be explained by the fact that the CT-WC interaction was highly centred on task-related issues. As a result, it is possible that students did not feel the need to communicate among them with regard to task procedures. As for student-teacher interaction, an alignment between some of the strategies used by the teachers (when addressing students) and the students (when addressing teachers) was found. As has already been mentioned, teachers seem to have very well defined separate roles. It might be that in line with these roles, students addressed teachers in different ways too. Regardless of the fact that students used metacognitive and social/affective strategies with both teachers, it is worth noting that they addressed the content teacher for task-related issues, whereas they addressed the language teacher for pedagogical reasons, coinciding with the strategies that teachers used with students in whole class/group situations. The fact that students used pedagogical strategies mainly when talking with the language teacher means that learners were able to identify that the role of the expert was embodied by the language teacher and not by the content one. They addressed the language teacher in pedagogical terms so as to co-construct knowledge with the help of the expert. Besides, students did not ask the content teacher about language issues. This goes very much in line with the role and the functions that the content teacher embodied, both in whole class and group interaction. Yet, students did not address her for pedagogical and organisational purposes but addressed the language teacher for these purposes. This might be connected with the fact that students used to address the language teacher

Explicit Plurilingualism in Co-taught CLIL …

211

more frequently than the content teacher. In addition, this may also be related to the results with respect to social/affective strategies, for the interaction of the students when addressing the language teacher was highly framed within social/affective strategies. This might have had an impact when dealing with pedagogical strategies and metacognitive strategies as well. It seems that students addressed the language teacher whenever they felt they had a doubt because they felt more comfortable when interacting with her than when interacting with the content teacher. Moreover, it can also be related to the fact that they perceived the language teacher as the expert who participated the most in conversations around construction of knowledge. Although students did address the content teacher for metacognitive issues, the number of instances with regard to the language teacher was always shown to be higher. The second research question was linked to the direction of interaction among participants. The analysis of classroom interaction revealed that the direction of interaction was mainly from students to teachers. This finding was not expected at all due to the given learning context, since CLIL settings should ideally promote peer interaction. This fact reveals great insights on the CLIL model that was implemented in this science course. As the interaction among students only represented a small part out of the total, it has been concluded that these science lessons were not framed within a truly communicative pedagogy. Some of the necessary elements that are claimed to reinforce communicative procedures, such as the availability of the L1 and the application of CLIL approaches were present in the classroom. However, as the interaction was predominantly teacher-centred, the outcome was not the desired one, since “exclusive whole-class discussions and teacher-centered interactions –frequently reported in immersion programs– have been identified as methods for modern language teaching that fail to produce desired results” (Devos, 2016, p. 12). The fact that the interaction was teacher-centred was analysed as a consequence of a lack of regulation of teachers’ responsibilities. A number of appropriate guidelines for managing effective co-teaching and L1 use should be provided to teachers in order to create meaningful communicative situations. This might also include L1 explicit instruction on appropriate language use so as to encourage the learners to take active control over their learning. Besides, the expertise of the students and the difficulty posed by the subject might have also played a role (Gené Gil et al., 2012) in diminishing peer interaction. Students were not familiarised with experimental procedures, so it is possible that they felt the need to address the teachers in a recurrent manner. This argument can be associated with the prominence attached to teacher-whole class interaction, for teachers might have also felt the need to constantly give instructions and clarify the objectives in the classroom setting. However, this issue can also be related to the fact that collaborative functions and management procedures were not well established or regulated. This might prompt a sense of unbalance in regard to classroom procedures on the teacher’s part. That is to say, it is possible that the teachers felt that something had not been clarified or explained thoroughly enough, leading to repeated interventions. This also correlates with the high percentages of content teacher’s interventions addressed to the whole-class. The content teacher regularly intervened addressing the whole group of students in order to translate what the language teacher said, just

212

I. Milán-Maillo and E. Pladevall-Ballester

in case students did not understand the message because of language difficulties. Nonetheless, the fact that the interaction among students just represented the 21.5% out of the total was found to be intriguing, for students had absolute freedom to use their L1 and always worked in teams. However, the arguments suggested above might provide an explanation, considering that teachers almost completely remained the centre of classroom interaction. Students could not interact among themselves more than they actually did, which led to an undesired outcome, for the main tenets of CLIL pedagogy were not fully developed. The fact that student-student interaction was that low in relation to the total complicated the integration and mitigated the effectiveness of L1 interactional strategies. As collaborative dialogue did not predominate, their learning process might have also been affected, as strategies and learning mediation mainly depend on peer and social interaction. The final section will close the chapter with a number of concluding remarks.

6 Conclusions The data presented in this paper has shown that the L1 acted as a compensatory strategy that coped with CLIL demands and was beneficial both for learners and teachers. Thus, this study enlarges the literature that provides evidence for the L1 being a beneficial tool in FL settings (Alley, 2005; Antón & DiCamilla, 1998; Arratibel-Irazusta & Martínez-Adrián, 2018; García Mayo & Hidalgo, 2017; Martínez-Adrián et al., 2017; Martínez-Adrián, 2018; Pavón & Ramos, 2018; Pladevall-Ballester & Vraciu, 2017; Storch & Aldosari, 2010; Swain & Lapkin, 2000; Tognini & Oliver, 2012). Although the idea that the L1 might be a damaging source of cross-linguistic influence when learning a foreign language is discarded in this study, L1 use did not always fulfil a cognitive strategic role. Yet, the native language will always be present for the foreign language speaker, and it will be used if the speaker feels the need to do so. Therefore, the potential benefits that a regularised use of the L1 might bring to foreign language instruction cannot be ignored. L1 use has to be focused and employed to the learners’ advantage so as to maximise the avoidance of using the L1 within unplanned discourse situations (Arratibel-Irazusta & Martínez-Adrián, 2018; Gené Gil et al., 2012). When restrictions are not applied on its use, the L1 does not fully develop its cognitive and scaffolding role within the classroom context. The cognitive support provided by the L1 is essential in the process of deriving meaningful FL learning (Alegría de la Colina & García Mayo, 2009). Taking this into account, a proper regulation of teacher’s responsibilities and a balance regarding the amount of L1 use is needed so as to create the optimal conditions to foster FL learning and maximise the potential functions the L1 can have during CLIL instruction (Pavón & Ramos, 2018). Maximising FL comprehensible input and learning opportunities alongside relying on L1 mediating functions and collaborative dialogue is necessary as well (Tognini & Oliver, 2012). If restrictions and regulations on language use were applied, the use of the L1 would probably be more centred on and related to learning and maintaining FL discourse during collaborative dialogue. This would be benefi-

Explicit Plurilingualism in Co-taught CLIL …

213

cial both for CLIL methodology and for CLIL learners, who would develop content and language learning on a higher cognitive basis (at least within communicative and cooperative situations). Besides, a regularisation of functions and a balance of teachers’ responsibilities are needed in order to effectively integrate collaborative teaching practices within the model, diminish the predominance of centred teaching and give more room for student-student interaction. This, which also fits CLIL communicative purposes better, would stimulate interactive group-work and peer scaffolding, which would produce more opportunities for cognitive and learning development. Effective and explicit guidelines need to be given to teachers so that the integration of content and language can be more easily handled during the lessons. In any case, it is clear that the primary objective is to truly apply a balanced content and language pedagogy, both concerning its methodology and its practice (Devos, 2016) so as to be able to establish a balanced collaborative teaching model. The potential offered by the L1 needs to be exploited so as to reveal more insights with regard to foreign language learning processes. In order to do so, planning the use of the L1 in FL settings is mandatory if strategic fostering of foreign language use is to be promoted. Acknowledgements We would like to express our gratitude to the school and the participants of this study, since it would not have been possible without their participation. We would also like to thank the EFLIC Research Group (2014SGR693, 2017SGR752), which also allowed us to collect the data for this study.

Native Language Identification by Human Judges Scott Jarvis, Rosa Alonso Alonso and Scott Crossley

Abstract This paper reports the results of two studies involving native language identification (NLI) by human judges (see Jarvis, 2012; Malmasi, Tetreault, & Dras, 2015; Odin, 1996). The first study includes six Finnish speakers who were asked to read a set of essays written in English and decide as quickly as possible whether each text was written by a Finnish or Swedish speaker. The second study includes the same six Finnish raters and an additional group of 10 Spanish-speaking raters. Both sets of raters were given 40 essays from the International Corpus of Learner English. The raters were asked to read each text carefully and indicate which texts they believed were written by native speakers of their own native languages. The results of both studies show that some of the raters were remarkably accurate in identifying the correct L1s of the writers, with accuracy rates above 80% for the top Finnish raters and over 90% for the top Spanish raters. A brief qualitative analysis of the data describes the types of features in the texts that allowed the raters to achieve such high accuracy rates. The paper concludes with a discussion of the theoretical, methodological, and pedagogical implications of these findings. Keywords Transfer · Cross-linguistic influence · Native language identification (NLI) · Human judgments

S. Jarvis (B) Department of Linguistics, University of Utah, 225 S Central Campus Drive Suite 2325, Salt Lake City, Utah 84112, USA e-mail: [email protected] R. Alonso Alonso Facultad de Filología y Traducción, Departamento de Filología Inglesa, Francesa y Alemana, Universidad de Vigo, As Lagoas, Marcosende s/n, Vigo, Spain e-mail: [email protected] S. Crossley Department of Applied Linguistics/ESL, Georgia State University, 25 Park Place, Suite 1500, Atlanta, GA 30303, USA e-mail: [email protected] © Springer Nature Switzerland AG 2019 M. J. Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2_11

215

216

S. Jarvis et al.

1 Introduction Native Language Identification (NLI) is a relatively new line of empirical exploration within the domain of transfer research. NLI developed gradually between 2001 and 2005 out of the converging interests of transfer researchers and computational linguists (Jarvis, 2012), and it has since burst into a vigorous area of exploration for researchers specializing in cross-linguistic influence (Jarvis & Crossley, 2012), learner corpus research (Jarvis & Paquot, 2015), and applied computational linguistics (Malmasi et al., 2017). Importantly, NLI is not so much a theoretical perspective or empirical discovery as it is a methodological approach. It refers to the use of machine-learning technology to classify samples of language use (e.g., samples of writing or transcriptions of speech) in accordance with the native languages (L1s) of the people who produced those samples. Researchers who conduct NLI studies make use of existing machine-learning tools—such as Linear Discriminant Analysis, Support Vector Machines, Logistic Regression, and so forth (Malmasi et al., 2017)—to create classification systems (or statistical models) designed to predict the correct L1 affiliation of any text that the classification system is applied to. The core of the system is a machine-learning tool whose parameters need to be set by the researcher. The rest of the system is determined by the researcher, and is made up of classes (i.e., categories) and features (i.e., variables) (Jarvis & Paquot, 2015). In NLI studies, classes are the L1s that the system is trying to learn to predict, and features are the measurable characteristics of texts that the system learns to associate with individual L1s. Some of the most commonly used features in past NLI studies have been the relative frequencies of occurrence (within each text) of specific words, lexical n-grams (e.g., combinations of two, three, etc. words), specific parts of speech (POS), POS n-grams, syntactic constructions, letters, letter n-grams, and so forth (Jarvis, Bestgen, & Pepper, 2013). NLI studies typically begin with a training phase where the classification system is given both the classes and features of each text comprising a set of training data. The system uses the training data to assign weights to each feature in such a way that the overall combination of features provides maximal separation between the classes (i.e., L1s) in the training data. Once the system has been trained in this way, it is then applied to a set of test data to see how accurately it can predict the correct classes of each text based on the features of those texts. Most NLI studies have demonstrated success in identifying the correct L1 affiliations of texts in their test sets—at levels significantly and often substantially above the level of chance. The purpose and value of conducting NLI research is not always the same for transfer researchers as it is for computational linguists. NLI studies carried out by computational linguists have generally focused on the classification accuracy of the system, the computational innovations that have given it its power, and the value of the system or its individual innovations for future research and future practical applications (Tetreault, Blanchard, Cahill, & Chodorow, 2012). NLI studies carried out by transfer researchers, on the other hand, have tended to focus on the discovery of which language features and combinations of features are distinctive of learners from

Native Language Identification by Human Judges

217

specific L1 backgrounds (Jarvis & Crossley, 2012). However, we acknowledge that research by computational linguists has also begun to move in this direction (Malmasi & Dras, 2014; Swanson & Charniak, 2014). Transfer researchers are also interested in the classification accuracy of the system insofar as it serves as an indication of how reliably different the language use of learners from different L1 backgrounds is—which also serves as a proxy for the prevalence of cross-linguistic influence in the data. However, the main goal of NLI research for transfer researchers is to discover previously overlooked ways in which elements of learners’ L1s carry over into their use of the target language (Jarvis, 2012). This same goal can be pursued by using human judges instead of (or in addition to) machine-learning technology, and, in fact, there are good reasons for doing so. Adult humans have large, rich, and highly contextualized knowledge bases that allow them to approach tasks with a great deal of power and flexibility. They can see things that machines cannot, and they can update their expectations in accordance with task demands and changes in context (such as differences in topic, genre, or the specific content of a text). Humans also take an open-ended approach to judgment tasks. Whereas machines consider only those features of texts that the researcher has programmed into the system, humans are sensitive to potentially any clue they might encounter in a text, including not only forms and structures, but also writers’ styles, rhetorical strategies, and various shades of meaning. Even though human judges’ open-ended approach can be a disadvantage in studies designed to focus on the effects of a narrow set of textual properties (i.e., since humans might not be capable of completely disregarding all other properties), their broad orientation can be an enormous advantage in exploratory studies designed to capture any of the subtle and contextualized forms of cross-linguistic influence that might exist in the data but might be unanticipated and would otherwise remain overlooked and undiscovered. The present paper deals with how well adult human judges who are highly proficient in a second language (L2) can single out which specific texts in a larger collection of L2 texts have been written by people who share their L1. The paper also deals with the types of human judges who are able to do this best, what types of strategies they use, and which specific language clues they rely on. Although it would be exciting to discover that most humans can do this, for the purposes of the present research even a single person who can perform L1 identification with an accuracy rate of, say, 80% or higher, would provide a valuable indication of the potential levels of L1 influence in the data and of how it manifests itself in texts whose L1 affiliations are correctly identified. The remainder of this chapter is organized as follows. Section 2 discusses past research on human native language identification, identifies important gaps in the existing literature, and presents the research questions that this paper will address. Section 3 reports the results of a study of the accuracy of human raters’ rapid, intuitive NLI judgments, whereas Sect. 4 reports a complementary study on the accuracy of human raters’ NLI judgments after careful scrutiny of L2 texts. A general discussion of the results and implications of both studies is provided in Sect. 5, and the chapter’s main findings and implications for future research are summarized in Sect. 6.

218

S. Jarvis et al.

2 Past Research There are only three previous studies we are aware of that have investigated whether people are capable of reading texts written in an L2 and detecting the L1 backgrounds of the writers of those texts. The first such study was conducted by Ioup (1984). The purpose of the study was to test whether native speakers of English could detect specific foreign accents in L2 English both with and without access to phonological cues. The data with phonological cues were taped speech samples produced in English by Arabic- and Korean-speaking learners. The data without phonological cues were texts written in English by learners whose native languages included Hebrew and Spanish. The results of the study showed that the English-speaking judges were indeed reliably able to detect the correct L1 affiliations of the speech samples but were not able to do so with the written texts at a level above chance. The author interpreted the results as showing that manifestations of syntactic—and by association also morphological and lexical—transfer are not strong enough or reliable enough to make L1 detection possible without access to phonological cues. However, Ioup’s (1984) interpretations have been challenged by the results of a study by Odlin (1996), who found that human judges are indeed capable of L1 detection involving written L2 data. Odlin presented bilingual judges who were proficient in either English and Korean or English and Spanish with sentences that had been produced by Korean- and Spanish-speaking learners of English. The sentences contained various types of structural errors, and the results showed that the bilingual judges were successful in detecting which sentences—and specifically which structural errors—were produced by learners whose L1 was one of the languages the bilingual judges knew. The results of this study thus show that syntactic—or at least a more general type of structural—cross-linguistic influence is indeed strong enough and reliable enough to make L1 detection possible in the absence of phonological cues. There are nevertheless some important differences between the Ioup (1984) study and the Odlin (1996) study. The most important difference is that Ioup asked native English-speaking judges to differentiate between the L2 patterns of learners whose L1s the judges apparently were not proficient in, whereas Odlin asked bilingual judges to identify the influence of an L1 that they were highly proficient in. The differential results of these studies suggest that human judges might be effective in performing L1 identification only with respect to those L1s in which they have advanced levels of knowledge. The third and final study we are aware of that has investigated NLI by human judges is Malmasi et al. (2015). The purpose of this study was to compare human raters against machine-learning systems in terms of their ability to identify the L1 affiliations of L2 English texts produced by writers from five very different L1 backgrounds. Unlike Odlin’s (1996) study, where the human judges were asked to pick out only those texts written by speakers of a single L1 in which the judges themselves were highly proficient, the judges in the Malmasi et al. study were asked to classify each of 30 texts as having been written by a native speaker of Arabic, Chinese, German, Hindi, or Spanish. In order to recruit human judges who had relevant

Native Language Identification by Human Judges

219

knowledge of and exposure to these languages, the researchers “chose 10 professors and researchers who have varied linguistic backgrounds, speak multiple languages, and have had exposure with the particular L1s, either as a speaker or through working with ESL students” (p. 175). The 30 texts given to the judges were selected from the TOEFL11 corpus. The results of the study showed that the human judges’ accuracy rates ranged from 30 to 53.3% (where the chance-level baseline was 20%), and that the machinelearning systems used for comparison outperformed all of the human judges by a wide margin. The researchers concluded that human judges are not likely to be able to perform nearly as well as machine-learning systems on NLI tasks involving multiple L1s. However, they acknowledged that this was only a pilot study, and that some of their human judges had little to no knowledge of most of the L1s in the study. The researchers also acknowledged that some of the judges had had very little exposure to L2 English written by speakers of Arabic, German, or Hindi. From our perspective, it would be useful to conduct a follow-up study where each of the judges is a native speaker of at least one of the L1s, and where the judges are asked to identify only the texts written by speakers of their own L1s. We believe that this would be a better test of what human judges are capable of with respect to NLI. In addition to these recommendations, we also think that it would be useful to explore how the NLI performance of human judges might complement rather than compete with the performance of machine-learning systems. Investigations of native language identification by human judges can be designed with varying purposes. The primary purpose of the studies by Ioup (1984) and Odlin (1996) was to determine the salience and pervasiveness of structural transfer. In both studies, the criterion for confirming the salience of structural transfer was whether human judges in general (i.e., as a group) could perform L1 identification successfully. This does indeed seem to be an appropriate criterion for evaluating salience. However, Odlin (1996, p. 175) cautioned that the salience of cross-linguistic influence has no bearing on whether such influence has actually occurred: “cross-linguistic influence can arise and not be salient at all”. From this, it follows that studies that make use of human judges but are not designed to address the question of salience have little reason to rely on aggregated group-level accuracy rates. Because high levels of L1 identification accuracy are impossible unless the data do in fact contain reliable indicators of the backgrounds of the people who produced the data, a high level of L1 identification by even a single human judge might serve as the basis for a successful, informative study of cross-linguistic influence. By analogy, we point to the fact that a single machine-learning system that is able to achieve high levels of L1 classification accuracy—even if all other systems fail—serves as strong evidence that reliable L1 indicators are pervasive in the data regardless of how subtle they might be. These important assumptions appear to have informed the study by Malmasi et al. (2015). A related, important purpose for conducting this type of research is, as we mentioned earlier, to discover which language features and combinations of features are distinctive of learners from specific L1 backgrounds, which in many cases might simultaneously lead to the discovery of previously overlooked ways in which elements of learners’ L1s carry over into their use of the target language. The

220

S. Jarvis et al.

two studies described in the following sections fill the gaps mentioned and address the following research questions: 1. How accurately can human judges identify the L1 affiliations of L2 texts written by learners who share their L1? a. Are there large disparities in the accuracy rates of judges who are most successful and least successful in performing L1 identification? b. Are there any judges who can identify L1s with more than 80% accuracy? c. What types of judges (e.g., analytical vs. holistic) are best able to perform L1 identification? d. Are there any observable differences in the strategies (e.g., efforts to identify specific types of L1 clues) that are used by judges who are the most and least successful in performing L1 identification? 2. Which textual features (or language-use cues) do successful judges rely on? a. Are these textual features indeed reliable indicators of learners’ L1s? b. Do these textual features reveal any previously unrecognized ways in which learners’ patterns of L1 use might carry over into their use of an L2? 3. What do the results suggest concerning the overall pervasiveness of L1 influence in the data, both within and across texts?

3 Study 1 3.1 Method The human judges in Study 1 consisted of a single group of six native Finnishspeaking students pursuing graduate degrees in the Department of Languages at the University of Jyväskylä, Finland. All six were females in their early to mid 20s, were highly proficient in English, and had also studied additional languages, including Swedish. One of the six, whom we will refer to as F1, had advanced knowledge of Swedish and was pursuing a graduate degree in the department’s Swedish language program. This is relevant because the task we gave to the judges was an L1 identification task that included texts written in L2 English by learners whose native languages were either Finnish or Swedish. The six human judges were given a collection of 16 narrative essays that were written retells of an eight-minute segment of a silent Charlie Chaplin film (Jarvis, 2000). Eight of the essays had been written by Finnish-speaking learners, and the remaining eight had been written by Swedish-speaking learners. The judges were not told how many essays represented each L1 background, and the essays were shuffled into a randomized order before being given to the judges. To avoid potential proficiency-related effects, the 16 essays that were selected for this study had all been rated at the mid-intermediate level of writing proficiency (or quality) (Jarvis, 2002). The texts ranged in length from 164

Native Language Identification by Human Judges

221

words to 341 words. The judges were given all 16 essays—each on a separate sheet of paper—and were asked to read through the essays as quickly as possible and to indicate whether they thought each essay was written by a Finnish speaker or Swedish speaker. Judge F1 completed the L1 identification task in less than five minutes, and the remaining judges completed it in under 15 min.

3.2 Results and Discussion The results of this task are shown in Table 1. The first column of the table shows the identification codes for each judge, and the next six columns show the recall, precision, and F1 values for each judge in relation to the texts written by Finnish speakers, on the one hand, and Swedish speakers, on the other. Recall values were calculated using a numerator representing the total number of texts identified correctly as either Finnish or Swedish, and using a denominator representing the total number of texts that actually were written by Finnish speakers or Swedish speakers (i.e., 8). Precision values were calculated by dividing the number of texts correctly identified as either Finnish or Swedish, by the number of texts judged to have been written by either Finns or Swedes. F1 values were calculated by dividing the product of the recall and precision values by the sum of the recall and precision values, and then by multiplying this quotient by 2. Importantly, recall values are sensitive to underidentifications and precision values are sensitive to over-identifications, whereas F1 values represent a balance between recall and precision. The final column, Overall Accuracy, shows the simple percentage of texts correctly categorized according to the L1 backgrounds of the learners who wrote them. Overall accuracy is often very similar to F1 , but overall accuracy can be over-inflated in cases where there is an uneven class distribution (especially when there is a large proportion of true negatives; see Manning, Raghavan, & Schütze, 2008).

Table 1 Native language identification by human judges: study 1 results Judge

Finnish recall (%)

Finnish precision (%)

Finnish F1 (%)

Swedish recall (%)

Swedish precision (%)

Swedish F1 (%)

Overall accuracy (%)

F1

75.00

85.70

79.99

87.50

77.80

82.37

81.30

F2

62.50

71.40

66.65

75.00

66.70

70.61

68.80

F3

75.00

66.70

70.61

62.50

71.40

66.65

68.80

F4

62.50

62.50

62.50

62.50

62.50

62.50

62.50

F5

62.50

71.40

66.65

62.50

62.50

62.50

62.50

50.00

50.00

50.00

50.00

50.00

50.00

50.00

F6 a The

chance-level baseline accuracy for this task is 50%

222

S. Jarvis et al.

The most noteworthy results in Table 1 are those of Judge F1. Her overall L1 identification accuracy of 81.30% surpasses the 80% threshold established in our research questions, and her recall and precision values show that she was successful in identifying nearly all (i.e., 7 out of 8) of the essays written by Swedish speakers while also identifying almost as many (i.e., 6 out of 8) of the essays written by Finnish speakers. Judges F2–F5 also showed overall accuracy rates well above chance, but these judges were nevertheless substantially less accurate than Judge F1. Judge F6 produced the lowest accuracy rates, performing exactly at the level of chance with respect to recall, precision, F1 , and overall accuracy. These results thus show that, as a general trend, human judges are indeed able to identify at levels above chance which texts in a collection of L2 texts were written by learners who speak the same L1 as they do. However, this is not true of all human judges, and the differences in accuracy rates between the least and most successful judges are substantial. These findings are generally in line with those of Malmasi et al. (2015). Importantly, in the present study, one of the judges surpassed the 80% accuracy threshold. This was a judge who—as mentioned earlier—was not only a native speaker of Finnish but was also a highly advanced speaker of Swedish. This suggests, as an extension to Odlin’s (1996) finding regarding the advantages of bilingual judges, that trilingual judges appear to be even more effective than bilinguals in an L1 identification task involving two L1s that the trilingual judge is highly proficient in. Although her high level of proficiency in both Finnish and Swedish appears to be the reason why Judge F1 had such a high accuracy rate and was able to complete the task so quickly, there may be an additional factor. Our observations of her work in other areas (including in Study 2) showed that she tends to perform most tasks quickly and without devoting much time to analysis or reflection. Her quick reading time indicates that she relies on intuition, and intuition served her well in this particular L1 identification task (but see the results of Study 2). For our purposes, however, the fact that she may have relied on intuition meant that she was not able to provide us with much information about which textual features she attended to when deciding whether each essay was written by a Finnish speaker or Swedish speaker. Nevertheless, she did indicate that the texts written by Swedish speakers tended to include misspellings that were indicative of Swedish influence (e.g., bagery ‘bakery’, Sw = bageri), and they also included an occasional phrase that was essentially a direct translation of a common Swedish expression (e.g., He did something ill ‘He did something bad’, Sw = Han gjorde nånting illa). It appeared that she made her L1 identification judgments based mainly on whether she found any traces of Swedish in the essays, rather than on whether they contained any traces of Finnish. She did not appear to rely on any previously undiscovered types of cross-linguistic influence, but it is important to note that she approached the task very differently than a machine-learning NLI system would have. Practically all existing (machinelearning) NLI studies rely on textual features that represent language-use patterns encountered numerous times across multiple texts in the L2 training data. Existing NLI systems are thus statistical learners that make their L1 identifications on the basis of quantitative considerations. Judge F1’s L1 identifications, on the other hand, appear to have been made mainly on the basis of qualitative information—on simi-

Native Language Identification by Human Judges

223

larities between what she saw in the L2 English data and what she knew was possible in the learners’ L1s (especially Swedish). She was therefore able to recognize L1 influence involving L2 forms and structures that (a) occurred in only one of the 16 essays, and (b) were things she had never encountered before in English. F1 is not necessarily better at L1 identification than a machine-learning system would be, but the fact that she achieved such a high level of identification accuracy while doing something very different from what a computer-automated system would do, suggests that successful human judges and successful machine-learning systems might complement each other in important ways. In other words, future research might benefit from using both at the same time. We will return to this point in Sect. 5. In the meantime, it is useful to point out that F1’s accuracy rate of 81.3% is somewhat higher than the 76.9% classification accuracy of a Linear Discriminant Analysis classifier applied to the same type of data in Jarvis, Castañeda-Jiménez, and Nielsen (2012). The results of the two analyses are not completely comparable, however, because the NLI task in Jarvis et al. required the machine-learning system to distinguish among five L1s (Danish, Finnish, Portuguese, Spanish, and Swedish) instead of just two, and it was also performed on the basis of the writers’ use of only 53 words without taking any other information into account. What we can say, based on the results of this study as well as that of Jarvis et al. (2012), is that the pervasiveness of L1 indicators in the Chaplin film retell data is probably at least 81%, if not higher. In other words, the results seem to show that reliable indicators of learners’ L1s appear to be present in at least 81% of the texts. The pervasiveness of L1 indicators within texts is apparently much lower than this, however, and in some cases the successful identification of the L1 affiliation of a text seems to have been made possible by the occurrence of only one L1 indicator within that text.

4 Study 2 4.1 Method The human judges in Study 2 included two groups of participants: the same Finnishspeaking judges who participated in Study 1 (with about a week between studies), and an additional group of 10 native speakers of Spanish. The Spanish-speaking judges were all teachers of English as a foreign language at the State Language School in Vigo, Spain. They were all Spanish-Galician bilinguals who were also highly proficient in English (i.e., they were highly-functioning multilinguals). The L1 identification task in this study included 40 argumentative essays written on a variety of topics that were extracted from the International Corpus of Learner English (ICLE, see Granger, Dagneaux, Meunier, & Paquot, 2009). The 40 essays included 10 essays written by learners from each of the following L1 groups: Czech, Finnish, German, and Spanish. Information about the proficiency levels of the learners who produced these essays was not available, but all essays included in ICLE represent

224

S. Jarvis et al.

upper-intermediate to advanced levels of writing proficiency (or quality). The number of words in each text ranged from 276 to 1159, with a mean length of 652.55 words and a standard deviation of 229.33 words. A one-way ANOVA with Tamhane post-hoc tests showed that the texts written by German speakers were significantly shorter than those written by each of the other groups, but there were no significant differences among the three other groups. The judges in both groups were given all 40 essays in one of two randomized orders. Two of the Finnish judges and five of the Spanish judges were given the essays in randomized order A, and the remaining four Finnish judges and five Spanish judges were given the essays in randomized order B. They were told that some of the essays in the batch were written by people who spoke the same native language they did, and they were asked to try to identify which essays they were. To help them do this, they were given a form that listed the name of each text (Text 1, Text 2,…Text 40), and they were asked to place an “x” on the form next to the name of each text they thought was written by a native speaker of their own native language. They were also asked to list specific reasons for their judgments. They were not given a specific time limit but were told that in some cases they might feel that they could make a judgment after looking at a text for only a minute or so, but in other cases they might want to spend several minutes examining the text more thoroughly. Both groups of judges spent an average of about one and a half hours on the task as a whole, and several of the judges in both groups commented that it was difficult and time-consuming.

4.2 Results and Discussion The L1 identification results for the Finnish judges are shown in Table 2, and the results for the Spanish judges are shown in Table 3. The L1 identification task in Study 2 was more challenging than the one in Study 1 because it involved a larger mix of L1s, a much lower chance of guessing correctly (only 25% in Study 2 vs. 50% in Study 1), and it included essays representing a higher level of English proficiency than those in Study 1. Crucially, higher-level texts can be expected to have fewer conspicuous L1 indicators. These factors were compensated only by the fact that the judges in Study 2 were given as much time as they needed to complete the task. The combination of these reasons might account for why the accuracy rates in Tables 2 and 3 have minimum values that are lower than—and maximum values that are higher than—those in Table 1. Interestingly, every judge in Study 2 had an accuracy rate higher than the level of chance. The Finnish judge with the lowest L1 identification accuracy in Study 2 is the same judge who had the lowest accuracy rate in Study 1 (Judge F6). This might indicate that this judge did not take the task seriously, or that she simply did not have the linguistic aptitude necessary to perform this type of task effectively. The Finnish judge with the highest accuracy in Study 2 (Judge 2, with an accuracy rate of 82.5%) is not the same person who exhibited the highest accuracy in Study 1 (Judge 1), and

Native Language Identification by Human Judges Table 2 Study 2 results—Finnish judges’ identification of texts written by Finnish speakers

Judge

Recall (%)

Precision (%)

F1 (%)

Accuracy (%)

F1

50.00

27.80

35.73

55.00

F2

80.00

61.50

69.54

82.50

F3

70.00

35.00

46.67

60.00

F4

20.00

28.60

23.54

67.50

F5

70.00

29.20

41.21

50.00

F6

60.00

25.00

35.29

32.50

a The

Table 3 Study 2 results—Spanish judges’ identification of texts written by Spanish speakers

225

chance-level baseline accuracy is 25%

Judge

Recall (%)

Precision (%)

F1 (%)

Accuracy (%)

S1

100.00

41.67

58.83

65.00

S2

80.00

33.33

47.06

55.00

S3

100.00

76.92

86.95

92.50

S4

100.00

41.67

58.83

65.00

S5

100.00

76.92

86.95

92.50

S6

100.00

33.33

50.00

50.00

S7

90.00

33.33

48.65

52.50

S8

100.00

31.25

47.62

45.00

S9

40.00

26.67

32.00

57.50

S10

40.00

36.36

38.09

67.50

a The

chance-level baseline accuracy is 25%

it is noteworthy that Judge 2 spent more time on the task (about 2.5 h) in Study 2 than did any of the other Finnish judges, whereas Judge 1 spent the least amount of time (about 30 min). As before, it appears that Finnish Judge 1 resorted largely to intuition while making her judgments, but intuition was far less effective in this task than it was in the previous task, probably because it involved higher-level essays, most of which were written by learners whose L1s she did not know. Finnish Judge 2, on the other hand, spent several minutes reading and re-reading each text and marking every feature she thought might reflect Finnish influence. This careful, analytical task approach served her well and allowed her to surpass the 80% accuracy threshold (although her F1 was only 69.54%). The L1-related features that she noted dealt mainly with phrases that are grammatical in English but which she nevertheless felt might reflect a Finnish-like way of expressing meaning. These include phrases such as if he would do it (pro if he would), doing this replacement (pro replacing X with Y ), and in the course of one evening (pro in just one evening). If these types of phrases do indeed reflect influence from Finnish, the influence is very subtle because they are not strictly erroneous in English and most of them also

226

S. Jarvis et al.

do not reflect literal, word-for-word translations from Finnish (in the course of one evening would be rendered in Finnish as yhden illan aikana = ‘one-GEN eveningGEN time-ESSIVE’). Judge F2 appeared to recognize how subtle and precarious these potential instances of Finnish influence were; she nearly always tried to find at least two such instances within an essay before judging it to have been written by a Finnish speaker. The most remarkable results in Study 2 relate to the two Spanish judges (S3 and S5) with L1 identification accuracy rates higher than 90% and F1 values close to 90% (Table 3). Both judges correctly identified all 10 texts written by Spanish speakers, and both over-identified only three texts that were not written by Spanish speakers. These judges did not provide many details about how they were able to identify texts as having been written by Spanish speakers, but of the texts they identified correctly, they listed the following features as having provided valuable clues: calques and Spanish-like expressions (they see it strange > ‘they view it as being strange’), choice of words (done > ‘made’; like we see > ‘as we see’), syntax (perhaps is more difficult), and article use (the society). All of these features except perhaps Spanish-like expressions seem to show that L1 indicators in the texts written by Spanish speakers might have been more conspicuous than those in the texts written by Finnish speakers. This might also account for why Judges S3 and S5 were able to achieve higher levels of L1 identification accuracy than Judge F2. On the other hand, unlike F2, Judges S3 and S5 were experienced teachers of English to speakers of their own native language, who see L2 English essays written by Spanish speakers nearly every day. Judges S3 and S5 therefore probably had a greater level of awareness of the range of forms, expressions, and constructions that learners of English who share their L1 tend to produce. It is enlightening to compare the L1 identification accuracies achieved in the present study with those of a parallel study by Crossley and McNamara (2012), who used a machine-learning system to identify L2 English texts written by speakers of the same four L1s: Czech, Finnish, German, and Spanish. Crossley and McNamara used the same corpus (ICLE) from which the texts in the present study were extracted but included 903 texts in their sample rather than just the 40 used in the present study. The machine-learning classifier used by the researchers was Linear Discriminant Analysis, and the features they fed into the system included 19 measures of cohesion, lexical sophistication, syntactic complexity, and conceptual knowledge. Their final cross-validated results showed that 14 out of the original 19 variables representing these categories turned out to be reliable discriminators of the L1 affiliations of the texts, which led to an overall accuracy of 65.8% (where chance was 25%), with 55.3% classification accuracy for the Finnish texts and 55.0% classification accuracy for the Spanish texts. It is interesting to note that the highest performing human judges in the present study showed accuracy rates considerably higher than these values, but it is also important to recognize the differences in the nature of the tasks. Most importantly, the human judges were asked to identify only those texts written by speakers of their own L1s, whereas the machine-learning system had to distinguish between influence from four separate L1s. Also, the human judges were allowed to consider any clues they might encounter in the data, whereas the machine-learning

Native Language Identification by Human Judges

227

system was set up to look only at a narrowly defined set of features. It is quite likely that the machine-learning system would have performed at a level more comparable to that of the human judges if it had been given access to a wider range of variables (Jarvis et al. 2013). For this and other reasons that we will discuss in the following section, it is probably best not to think of either human judges or machine-learning systems as being superior to each other, but instead to consider the important ways they might complement each other.

5 General Discussion Our first research question deals with how accurately human judges are able to identify the L1 affiliations of L2 texts written by learners who speak the same L1s as the judges, whether there are large differences in the accuracy rates of the most and least successful judges, whether any judges are able to achieve an accuracy rate above 80%, and, if so, which types of people are able to do this and what strategies they use. The results of both studies and both sets of judges show that human judges who are highly proficient in the target language tend to be able to identify texts written by speakers of their own L1s at levels above chance. There was only one participant (F6 in Study 1) whose L1 identification accuracy was not above the level of chance, and only one additional participant (F4 in Study 2) whose F1 value was below the level of chance. The differences between the accuracy rates of those who do best versus worst on this type of task are quite large (82.5 vs. 32.5% among the Finnish judges in Study 2). Importantly, some judges are able to surpass 80% accuracy—or even 90% in the case of two Spanish judges in Study 2. It is not yet clear whether there is a particular profile for judges who are able to do this, but the findings of the present investigation combined with those of Ioup (1984), Odlin (1996), and Malmasi et al. (2015) suggest that a necessary prerequisite is advanced knowledge of the target language and of at least one of the native languages of the writers. Judge F1’s performance in Study 1 additionally suggests that advanced proficiency in more than one of the native languages of the writers further enhances L1 identification accuracy. The high level of accuracy achieved by Judges S3 and S5 in Study 2 suggests that experience teaching the target language to speakers of at least one of the L1s can also be beneficial, although this does not guarantee high levels of L1 identification accuracy, given that all of the Spanish judges were English teachers, but only two of them achieved L1 identification accuracy above 80%. Importantly, judges’ own experiences learning and/or teaching the language likely provide them with heightened levels of metalinguistic awareness of the L1, the L2, the relationship between the two languages, and the types of difficulties a speaker of the L1 encounters while learning the L2. According to Odlin (1996), these types of metalinguistic awareness might be critical to human judges’ ability to recognize instances of transfer. We have not been able to identify any strategies shared by all of the judges whose L1 identification accuracy rates exceeded 80%. The most successful strategies

228

S. Jarvis et al.

seemed to depend on the precise nature of the task and the compatibility between the task and the judge’s own strengths and proclivities. Judge F1’s intuitive approach worked well in Study 1, where (a) the judges were under time pressure, (b) the essays were written by speakers of only two L1s, (c) the judge was highly proficient in both L1s, and (d) the writers of the essays were at a level of L2 proficiency where instances of L1 influence are relatively conspicuous. None of these conditions held for the task in Study 2, however, and F1’s intuitive approach did not work well at all in this study. What did work well in Study 2 was Judge F2’s careful and comprehensive review of each text and her analysis of potential similarities—even abstract similarities—between acceptable but somewhat peculiar phrases in the L2 and the ways in which the same meanings would be expressed in L1 Finnish. Her approach was that of a veritable language detective, and she was careful not to reach a decision about the L1 affiliation of a text until she had found more than one likely L1 indicator per text. Unfortunately, the strategies used by the Spanish judges are much less clear, as are the details of how they arrived at their decisions about which texts were written by speakers of their own L1. Our second research question deals with which textual features (or language cues) the most successful judges rely on, and whether these cues represent any previously unrecognized ways in which learners’ patterns of L1 use might carry over into their use of an L2. The two most successful Spanish judges in Study 2 reported that the types of clues that helped them the most in their attempts to identify influence from Spanish included individual word choices, the use of characteristically Spanish expressions, word-for-word translations from Spanish, use of articles, and word order. Several of these types of clues—especially the last three—are structural in nature, and the fact that the Spanish judges found them to be so helpful confirms Odlin’s (1996) assertion that structural L1 transfer is indeed salient enough and pervasive enough to allow human judges to perform L1 identification successfully. The Spanish judges’ additional reliance on Spanish-like expressions coincides with the major type of clue that Judge F2 relied on in Study 2: phrases that reflect Finnish-like ways of expressing meaning even when such expressions are also acceptable in English (in the course of one evening). The influence from Finnish in these phrases is both indirect and subtle. One might even question whether these phrases really do reflect influence from Finnish. The evidence that they do is found in how reliably they led Judge F2 to the correct L1 identifications. Importantly, most of the L1 indicators that F2 highlighted and relied on so successfully are not instances of transfer that have been documented in previous research and are also so infrequent in the data that a machine-learning system would probably not make use of them. F2’s discoveries are thus valuable fodder for future investigations of L1 influence. More generally, explanations given by successful human judges like F2 might lead to useful new ways of engineering machine-learning features and systems in the future. Our third research question addresses the overall pervasiveness of L1 influence in the data, both within and across texts. A useful indicator of the pervasiveness of L1 influence across texts is the highest L1 identification accuracy achieved by any of the judges, which was 81.3% in Study 1, 82.5% for the Finnish texts in Study 2, and 92.5% for the Spanish texts in Study 2. We believe that these figures provide a fairly

Native Language Identification by Human Judges

229

good estimate of the proportion of texts in the present data that contain language use that is distinctive of specific L1 backgrounds. It may even be that the highest accuracy rates found in the present investigation are actually underestimates of the pervasiveness of L1 influence in the data, at least in terms of the proportion of texts that contain any detectable traces of the L1. Even though the best human judges did remarkably well, it is likely that not even they recognized every vestige of L1 influence embedded in the data. Nevertheless, for present purposes, we will assume that the overall pervasiveness of L1 influence in the data is approximately equal to the highest levels of L1 identification attained by any of the judges. Crucially, such high levels of L1 identification are not possible in the absence of reliable cues that relate texts to the groups of people who produced those texts. Although it is possible for the data to contain other types of cues that are not manifestations of L1 influence but which nonetheless distinguish texts written by learners from differing backgrounds—cues such as differences in text length, levels of proficiency, references to the writers’ home countries, learned stylistic conventions (e.g., the use of punctuation), other effects of teaching (e.g., L2 phrases learned from their textbooks or teachers)—we did our best to eliminate these types of cues from the data. We appear to have been successful because the human judges did not report finding any such cues in the texts they were asked to evaluate. The pervasiveness of L1 influence within texts is more difficult to assess, but on the surface it seems to be much lower than the pervasiveness across texts. Evidence for this is found in the fact that the judges sometimes identified the correct L1 affiliation of a text after finding only one or two instances of language use implicating a particular L1. In other words, L1 influence in many cases might be limited to a single morpheme, word, or phrase within a text. On the other hand, there might be other types of L1 influence that are more implicit and more pervasive than this. To see this better, it is useful to compare the results of the present investigation with those of parallel studies that have applied machine-learning systems to the corpora from which the texts in the present investigation were extracted. As mentioned previously, the present investigation with human judges rendered higher levels of L1 identification accuracy than the corresponding machine-learning studies (81.3% in Study 1 vs. 76.9% in Jarvis et al., 2012; 82.5% for the Finnish texts in Study 2 vs. 55.3% for the Finnish texts in Crossley and McNamara, 2012; 92.5% for the Spanish texts in Study 2 vs. 55.0% for the Spanish texts in Crossley and McNamara, 2012). We acknowledge that the results of the different studies are not fully comparable because of major differences in the nature of the tasks that the human judges versus machines were given, but comparing the results of the two types of studies highlights some valuable points: First, the best human judges appear to be better than machines at identifying the pervasiveness of L1 influence across texts (but see Malmasi et al., 2015). Second, the best human judges are able to do this on the basis of as few as one or two L1 indicators per text. Third, the L1 indicators that the best human judges rely on are sometimes exceedingly rare in the data. Fourth, the L1 indicators they rely on reflect at least a subtle cross-language congruity between the L1 and the writer’s use of the L2, but because they are often rare, they tend not to represent within-group similarities

230

S. Jarvis et al.

or between-group differences1 (Jarvis, 2000), which are fundamental to machinelearning systems (Jarvis, 2012). Finally, machine-learning systems are sensitive to subtle statistical variances in numerous features found throughout each text. These findings illustrate that, whereas the most successful human judges were able to identify L1 influence in more texts, the machine-learning systems—when accurate—were able to identify L1 influence in a larger number of features within each text. Human judges and machine-learning systems clearly do not follow the same approach to L1 identification, but the combined results of both seems to point to a fairly high pervasiveness of L1 influence both within and across texts. Future research would clearly benefit from using a combination of human judges and machine-learning systems to probe these questions further, and especially to determine whether humans and machines identify the correct L1 affiliations of the same texts, and whether they are also capable of recognizing L1 influence in the same textual features. If not, it might be useful for future NLI studies to arrive at their final L1 classifications based on a combination of the predictions of both human judges and machine classifiers. As pointed out by Malmasi et al. (2017), the currently most effective NLI systems are multiple classifier systems (also called ensembles and meta-classifiers), which combine the predictions of multiple machine-learning systems in order to arrive at their final L1 identification decisions. Including human judgments within these multiple classifier systems might be an important step forward in gauging the true impact of L1 influence and uncovering all of its nuanced dimensions. The results of the present investigation do not offer straightforward pedagogical implications. However, we believe that this area of research will ultimately lead to detailed, nuanced information about the ways in which learners’ L1 knowledge and language-use patterns tend to influence and even carry over into their use of another language. Language learners will find this type of information valuable because it can raise their awareness of the forces that guide their own language choices and set them apart from speakers of other languages. We believe that this type of information will also empower language learners to use the target language more flexibly and in ways that produce the effects learners are hoping for. Some of the pedagogical benefits of this area of research will undoubtedly have to wait for the findings of future studies. In the meantime, however, we believe that it might be beneficial for language teachers to have their students engage in L1 identification tasks. This will raise their awareness of the ways in which L1 influence manifests itself, and if the L1 identification task includes samples of language use that the learners themselves have produced, then this type of exercise will also provide them with valuable feedback on the ways in which their own backgrounds have predisposed them to use the target language in specific ways. This could be accompanied with useful discussions about 1 As explained by Jarvis (2000), cross-language congruity (or L1-IL congruity) refers to similarities

between learners’ use of their L1 and the target language. Within-group similarities (or intra-group homogeneity) refer to similarities in the use of the target language by speakers of the same L1. Between-group differences (or inter-group heterogeneity) refer to differences in the use of the target language by speakers of different L1s. Jarvis argues that these are the three fundamental types of evidence for the presence of crosslinguistic influence in a set of learner data.

Native Language Identification by Human Judges

231

the positive, negative, and neutral effects of these predispositions, and of alternatives in language use (e.g., alternative options for expressing specific meanings) that might allow them to use the target language more flexibly and with more desirable effect.

6 Conclusion The results of the present study show that at least some human judges are capable of high levels of accuracy in the identification of L2 texts that have been written by language learners whose L1 is the same as that of the judges. This confirms the findings of Odlin (1996) and might be what Malmasi et al. (2015) would have found if they had asked their judges to identify texts written only by learners who shared the judges’ L1s. Our study also has important implications for the pervasiveness of L1 influence—including structural L1 influence—both within and across texts in a corpus. The nature of the results suggests that future NLI research might benefit from combinations of human judges and machine-learning systems for purposes of gauging the extent of L1 influence within and across texts, determining the precise nature of that influence, and discovering types of L1 influence that have been overlooked in past research. Although the pedagogical implications of this research are still unclear, we have suggested that language learners might benefit from engaging in L1 identification tasks themselves in order to become more aware of the forces that bind them as language learners and to become more capable of breaking free from such forces. Above all, the findings of the present investigation and related studies involving both human and computer-automated L1 identification show that there is still a good deal to be discovered about cross-linguistic influence.

References

Abutalebi, J., & Green, D. (2007). Bilingual language production: The neurocognition of language representation and control. Journal of Neurolinguistics, 20, 242–275. Achiba, M. (2003). Learning to request in a second language: A study of child interlanguage pragmatics. Clevedon: Multilingual Matters. Agustín-Llach, M. P. (2009). The role of Spanish L1 in the vocabulary use of CLIL and non-CLIL EFL learners. In Y. Ruiz de Zarobe & R. M. Jiménez Catalán (Eds.), Content and language integrated learning: Evidence from research in Europe (pp. 112–129). Bristol: Multilingual Matters. Agustín Llach, M. P. (2011). Lexical errors and accuracy in foreign language writing. Bristol: Multilingual Matters. Alcaraz, E., & Moody, B. (1999). Fonética inglesa para españoles. Alcoy: Marfil. Alcón, E. (2008). Investigating pragmatic language learning in foreign language classrooms. International Review of Applied Linguistics in Language Teaching, 46(3), 173–195. Alegría de la Colina, A., & García Mayo, M. P. (2009). Oral interaction in task-based EFL learning: The use of the L1 as a cognitive tool. IRAL-International Review of Applied Linguistics in Language Teaching, 47(3–4), 325–345. Aliaga, C. (2007). The role of phonetic training in L2 speech learning. In Proceedings of the Phonetics Teaching and Learning Conference (PTLC2007). http://www.phon.ucl.ac.uk/ptlc/ ptlc2007_web_procindex.html. Accessed April 04, 2018. Aliaga, C. (2009). Effects of audiovisual auditory versus articulatory training on L2 vowel category learning. In Proceedings of the Phonetics Teaching and Learning Conference (PTLC2009).http://www.phon.ucl.ac.uk/ptlc/ptlc2009/proceedings-index.html. Accessed April 04, 2018. Aliaga-Garcia, C., Joan, C. M., & Cerviño-Povedano, E. (2010). Phonological short-term memory and L2 speech learning in adulthood. In Proccedings of the 6th International Symposium on the Acquisition of Second Language Speech: New Sounds, 2010 (pp. 19–24). Allami, H., & Nekouzadeh, M. (2011). Congratulations and positive politeness strategies in Iranian Context. Theory and Practice in Language Studies, 1(11), 1607–1613. Alley, D. C. (2005). A study of Spanish II high school students’ discourse during group work. Foreign Language Annals, 38(2), 250–258. Allwright, R. L. (1996). Making sense of life in the language classroom: The significance of participant orientations. ESP Malaysia, 4, 41–43.

© Springer Nature Switzerland AG 2019 M. J Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2

233

234

References

Alonso Gonzalez, J., & Rothman, J. (2017). From theory to practice in multilingualism: What theoretical research implies for third language learning. In T. Angelovska & A. Hahn (Eds.), L3 syntactic transfer: Models, new developments and implications (pp. 277–298). Amsterdam: John Benjamins. Anderson, R. (1983). Transfer to somewhere. In S. M. Gass & L. Selinker (Eds.), Language transfer in language learning (pp. 177–201). Rowley, MA: Newbury House. Angelovska, T., & Hahn, A. (2012). Written L3 (English): Transfer phenomena of L2 (German) lexical and syntactic properties. In D. Gabrys-Barker (Ed.), Cross-linguistic influences in multilingual language acquisition, second language learning and teaching (pp. 23–41). Springer-Verlag Berlin Heidelberg. https://doi.org/10.1007/978-3-642-29557-7_2. Angelovska, T., & Hahn, A. (Eds.). (2017). L3 syntactic transfer: Models, new developments and implications. Amsterdam/Philadelphia: John Benjamins Publishing Company. Antón, M., & DiCamilla, F. (1998a). Socio-cognitive functions of L1 collaborative interaction in the L2 classroom. Canadian Modern Language Review, 54(3), 314–342. Antón, M., & DiCamilla, F. J. (1998b). Socio-cognitive functions of L1 collaborative interaction in the L2 classroom. The Modern Language Journal, 83(2), 233–246. [Word order in modern Apridonidze, S. (1986). Georgian]. Tbilisi: Mecniereba. Arabski, J. (2006). Cross-linguistic influences in the second language lexicon. Clevedon: Multilingual Matters. Arnfast, J. S., & Jorgensen, J. (2003). Code-switching as a communication, learning, and social negotiation strategy in first-year learners of Danish. International Journal of Applied Linguistics, 13(1), 23–53. Aronin, L., Fishman, J. A., Singleton, D., & O’Laoire, M. (2013). Current multilingualism: A new linguistic dispensation. In D. Singleton, J. A. Fishman, L. Aronin, & M. O’Laoire (Eds.), Current multilingualism (pp. 3–23). Berlin: Mouton/De Gruyter. Aronson, H. I. (1982). Georgian: A reading grammar. Chicago: Slavica. Arratibel-Irazusta, I., & Martínez-Adrián, M. (2018). The use of previously known languages in the oral production of L3 English learners: A pseudolongitudinal study. European Journal of Applied Linguistics, 6(2), 229–254. https://doi.org/10.1515/eujal-2017-0001. Auer, E. T. (2009). Spoken word recognition by eye. Scandinavian Journal of Psychology, 50(5), 419–425. https://doi.org/10.1111/j.1467-9450.2009.00751.x. Auer, P., & Wei, L. (2007). Introduction: Multilingualism as a problem? Monolingualism as a problem? In P. Auer & L. Wei (Eds.), Handbook of multilingualism and multilingual communication (pp. 1–13). Berlin: Mouton/De Gruyter. Austin, J. L. (1962). How to do things with words. Harvard: Harvard University Press. Avello, P., Lara, R., Mora, J. C, & Pérez-Vidal, C. (2013). The impact of study abroad and length of study on phonological development in speech production. In À. Llanes, L. A Ciró, L. Gallego, & R. Mateus (Eds.), Applied linguistics in an age of globalisation (pp. 67–73). Lérida: Ediciones de la Universidad de Lérida. Avello, P., Mora, J. C., & Pérez-Vidal, C. (2012). Perception of FA by non-native listeners in a study abroad context. Research in Language, 10(1), 63–78. https://doi.org/10.2478/v10015011-0050-9. Avery, P., & Ehrlich, S. (1992). Teaching American English pronunciation. Oxford: Oxford University Press. Ávila-Muñoz, A. M., & Sánchez-Sáez, J. M. (2014). Fuzzy sets and prototype theory. Representational model of cognitive community structures based on lexical availability trials. Review of Cognitive Linguistics, 12(1), 133–159. Azkarai, A., & García Mayo, M. P. (2017). Task repetition effects on L1 use in EFL child task-based interaction. Language Teaching Research, 21(4), 480–495. Azkarai, A., & Imaz Agirre, A. (2016). Negotiation of meaning strategies in child EFL mainstream and CLIL settings. TESOL Quarterly, 50(4), 844–870.

References

235

Bailey, N., Madden, C., & Krashen, S. (1974). Is there a natural sequence in adult second language learning? Language Learning, 24, 235–243. Ballesteros Martín, F. J. (2001). La cortesía española frente a la cortesía inglesa. Estudio pragmalingüístico de las exhortaciones impositivas. Estudios Ingleses de la Universidad Complutense, 9, 171–207. Bardel, C. (2006). La connaissance d’une langue étrangère romane favorise-t-elle l’acquisition d’une autre langue romane? Influences translinguistiques dans la syntaxe en L3. Acquisition et Interaction en Langue Étrangère, AILE, 24, 149–179. Bardel, C., Erickson, G., Granfeldt, J., & Rosén, C. (2017). Offering research education for in-service language teachers. Language Teaching, 50, 290–293. Bardel, C., & Falk, Y. (2007). The role of the second language in third language acquisition: The case of Germanic syntax. Second Language Research, 23, 459–484. Bardel, C., & Falk, Y. (2012). The L2 status factor and the declarative/procedural distinction. In J. Cabrelli Amaro, S. Flynn, & J. Rothman (Eds.), Third language acquisition in adulthood (pp. 61–78). Amsterdam: John Benjamins. Bardel, C., Falk, Y., & Lindqvist, C. (2013). Multilingualism in Sweden. In D. Singleton, J. Fishman, L. Aronin, & M. Ó Laoire (Eds.), Current multilingualism. A new linguistic dispensation (pp. 247–269). Boston/Berlin: Mouton De Gruyter. Bardel, C., & Sánchez, L. (2017). The L2 status factor hypothesis revisited: The role of metalinguistic knowledge, working memory, attention and noticing in third language learning. In T. Angelovska & A. Hahn (Eds.), L3 syntactic transfer: Models, new developments and implications (pp. 85–101). Amsterdam: Benjamins. Bardovi-Harlig, K. (2013). Developing L2 pragmatics. Language Learning, 63(1), 68–86. Bardovi-Harlig, K., & Vellenga, H. E. (2012). The effect of instruction on conventional expressions in L2 pragmatics. System, 40, 77–89. Barón, J., & Celaya, M. L. (2010). Developing pragmatic fluency in an EFL context. Eurosla Yearbook, 10(1), 38–61. Barriuso, T. A., & Hayes-Harb, R. (2018). High variability phonetic training as a bridge from research to practice. CATESOL Journal, 30(1), 177–194. Bataineh, R. F. (2014). On congratulating, thanking, and apologizing in Jordanian Arabic and American English. Journal of Intercultural Communication, 32, 1404–1634. Bell, A. (1984). Language style as audience design. Language in Society, 13(2), 145–204. Berkes, E., & Flynn, S. (2012). Further evidence in support of the cumulative-enhancement model. In J. Cabrelli Amaro, S. Flynn, & J. Rothman (Eds.), Third language acquisition in adulthood (pp. 143–164). Amsterdam: John Benjamins. Berry, V. (1997). Gender and personality as factors of interlocutor variability in oral performance tests. Paper presented at 19th Language Testing Research Colloquium, Orlando, Florida. Best, C. T., & Tyler, M. D. (2007). Non-native and second-language speech perception: Commonalities and complementarities. In O.-S. Bohn & M. J. Munro (Eds.), Language experience in second-language speech learning: In honor of James Emil Flege (pp. 13–34). Amsterdam: John Benjamins. Bialystok, E. (1983). Some factors in the selection and implementation of communication strategies. In C. Faerch & G. Kasper (Eds.), Strategies in interlanguage communication (pp. 100–118). London: Longman. Bialystok, E. (1991). Metalinguistic dimensions of bilingual language proficiency. In E. Bialystok (Ed.), Language processing in bilingual children (pp. 113–140). Cambridge, UK: Cambridge University Press. Bialyistok, E. (2001). Bilingualism in development: Language, literacy and cognition. New York: Cambridge University Press. Bialystok, E., & Barac, R. (2012). Emerging bilingualism: Dissociating advantages for metalinguistic awareness and executive control. Cognition, 122(1), 67–73.

236

References

Bijvoet, E., & Fraurud, K. (2010). ‘Rinkeby Swedish’ in the mind of the beholder. Studying listener perceptions of language variation in multilingual Stockholm. In P. Quist & B. A. Svendsen (Eds.), Multilingual urban Scandinavia. New linguistic practices (pp. 170–188). Bristol/Buffalo/Toronto: Multilingual Matters. Blum-Kulka, S., House, J., & Kasper, G. (Eds.). (1989). Cross-cultural pragmatics: Requests and apologies. Norwood, NJ: Albex. Boeder, W. (2005). The South Caucasian languages. Lingua, 115, 5–89. Boersma, P. (1998). Functional phonology. Doctoral dissertation, University of Amsterdam. The Hague: Holland Academic Graphics. Boersma, P., & Escudero, P. (2008). Learning to perceive a smaller L2 vowel inventory: An Optimality Theory account. In P. Avery, B. E. Dresher, & K. Rice (Eds.), Contrast in phonology: Theory, perception, acquisition (pp. 271–301). Berlin: Mouton de Gruyter. Boersma, P., Escudero, P., & Hayes, R. (2003, August). Learning abstract phonological from auditory phonetic categories: An integrated model for the acquisition of language-specific sound categories. In M. J. Solé, D. Recasens, & J. Romero (Eds.), Proceedings of the 15th International Congress of Phonetic Sciences (pp. 1013–1016). Barcelona: Universitat Autonoma de Barcelona. Bohnacker, U. (2006). When Swedes begin to learn German. From V2 to V2. Second Language Research, 22, 443–486. Borg, S. (2015). Teacher cognition and language education. Research and practice. London: Bloomsbury. Bouvy, C. (2000). Towards the construction of a theory of cross-linguistic transfer. In J. Cenoz & U. Jessner (Eds.), English in Europe. The acquisition of a third language (pp. 143–156). Clevedon: Multilingual Matters. Boyd, S., & Huss, L. (2001). Managing multilingualism in the traditional nation-state: The challenge for Sweden. Clevedon, UK: Multilingual Matters. Broner, M. A. (2001). Impact of interlocutor and task on first and second language use in a Spanish immersion program (Vol. 18). Center for Advanced Research on Language Acquisition, University of Minnesota. Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in linguistic usage. Cambridge: Cambridge University Press. Bruhn de Garavito, J., & Perpiñán, S. (2014). Subject pronouns and clitics in the Spanish interlanguage of French L1 speakers. In Proceedings of the 2014 annual conference of the Canadian Linguistic Association. http://cla-acl.ca/wp-content/uploads/Bruhn_Perpiñán-2014. pdf. Buckingham, A. (1997). Oral language testing: Do the age, status and gender of the interlocutor make a difference. Unpublished M.A. dissertation, University of Reading. Calvo-Benzies, Y. J. (2011). The teaching and learning of the pronunciation of English in Spanish secondary schools and universities. A preliminary analysis. Unpublished master´s thesis. Universidade de Santiago de Compostela, Santiago de Compostela, Spain. Calvo-Benzies, Y. J. (2013). Advanced Spanish university students’ problems with the pronunication of English vowels. Identification, analysis and teaching implications. Odisea Revista de Estudios Ingleses, 14, 37–56. Calvo-Benzies, Y. J. (2014). He was /gǝʊɪn/ to have a /bæɵ/, twenty /pɒrsent/ of /pɪpel daʊnlɒd / mʊsɪk/. A preliminary study of the difficulties shown by Spanish students of different levels of proficiency in the learning of English pronunciation. In P. Rodríguez-Puente, T. Fanego, E. Gandón, S. Riveiro, & M. L. Roca (Eds.), Current research in applied linguistics: Issues on language and cognition (pp. 124–153). Newcastle-upon-Tyne: Cambridge Scholars Publishing. Calvo-Benzies, Y. J. (2016). The teaching and learning of English pronunciation in Spain. An analysis and appraisal of students’ and teachers’ views and teaching materials. Unpublished doctoral dissertation. Universidade de Santiago de Compostela, Santiago de Compostela.

Carcedo González, A. (2000). Disponibilidad léxica en español como lengua extranjera: el caso finlandés (estudio del nivel preuniversitario y cotejo con tres fases de adquisición). Turku: Universidad de Turku. Carlisle, R. (1998). The acquisition of onsets in a markedness relationship. Studies in Second Language Acquisition, 20(2), 245–260. Castellotti, V. (2001). La langue maternelle en classe de langue étrangère. Paris: CLE international. Cebrian, J. (2009). Exploring the roles of instruction and word familiarity in L2 vowel identification. In Proceedings of the phonetics teaching and learning conference (PTLC2009). Retrieved from http://www.phon.ucl.ac.uk/ptlc/ptlc2009/proceedings-index.html. Accessed April 04, 2018. Cebrian, J., & Carlet, A. (2012). Audiovisual perception of native and non-native sounds by native and non-native speakers. In S. Martín-Alegre, M. Moyer, E. Pladevall, & S. Tubau (Eds.), At a time of crisis: English and American studies in Spain (pp. 300–307). Barcelona: Departament de Filologia Anglesa i de Germanística, Universitat Autònoma de Barcelona. Cebrian, J., & Carlet, A. (2014). Second-language learners´ identification of target language phonemes: A short term phonetic training study. Canadian Modern Language Review, 70(4), 474–499. https://doi.org/10.3138/cmlr.2318. Celaya, M. L. (1992). Present continuous or simple present?: Mother tongue influence in EFL. Atlantis, 14, 29–36. Celaya, M. L. (2008). “I study natus in English”: Lexical transfer in CLIL and regular learners. In R. Monroy & A. Sánchez (Eds.), 25 Años de lingüística aplicada en España: Hitos y retos (pp. 43–49). Murcia: Editum (Ediciones de la Universidad de Murcia). Celaya, M. L., & Barón, J. (2015). The interface between grammar and pragmatics in EFL measurement and development. European Journal of Applied Linguistics, 3(2), 181–203. Celaya, M. L., & Ruiz de Zarobe, Y. (2008). CLIL, age, and L1 influence. Paper presented at the XXXII AEDEAN Conference. University of the Balearic Islands, Palma de Mallorca, 13–15, November. Celaya, M. L., & Ruiz de Zarobe, Y. (2010). First language and age in CLIL and non-CLIL contexts. International CLIL Research Journal, 1(3), 60–66. Celaya, M. L., & Torras, M. R. (2001). L1 influence and EFL vocabulary. Do children rely more on L1 than adult learners? In Proceedings of the 25th AEDEAN Conference (pp. 1–14). Granada: Universidad de Granada. Cenoz, J. (2001). The effect of linguistic distance, L2 Status and age on cross-linguistic influence in third language acquisition. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Cross-linguistic influence in third language acquisition: Psycholinguistic approach (pp. 8–20). Clevedon: Multilingual Matters. Cenoz, J. (2003a). Cross-linguistic influence in third language acquisition: Implications for the organization of the multilingual mental lexicon. Bulletin Suisse de Linguistique Appliquée, 78, 1–11. Cenoz, J. (2003b). Cross-linguistic influence in third language acquisition: Implications for the organization of the multilingual mental lexicon. Bulletin VALS-ASLA (Vereinigung für angewandte Linguistik in der Schweiz), 78, 1–11. Cenoz, J. (2003c). The additive effect of bilingualism on third language acquisition: A review. International Journal of Bilingualism, 7(1), 71–78. Cenoz, J. (2009). Towards multilingual education: Basque educational research from an international perspective. Clevedon: Multilingual Matters. Cenoz, J., & García-Lecumberri, M. L. (1997a). The acquisition of English pronunciation: Learners’ views. International Journal of Applied Linguistics, 9, 3–18. Cenoz, J., & García-Lecumberri, M. L. (1997b). L2 perception of English vowels: Testing the validity of Kuhlʼs prototypes. Revista Alicantina de Estudios Ingleses, 10, 55–68. https://doi. org/10.14198/raei.1997.10.06.

238

References

Cenoz, J., & García-Lecumberri, M. L. (1999). The effect of training on the discrimination of English vowels. IRAL, 37(4), 261–275. https://doi.org/10.1515/iral.1999.37.4.261. Cenoz, J., & Jessner, U. (2000). English in Europe. The acquisition of a third language. Clevedon, UK: Multilingual Matters. Cenoz, J., & Valencia, J. (1994). Additive trilingualism: Evidence from the Basque Country. Applied Psycholinguistics, 15, 195–207. Cerviño, E., & Mora, J. C. (2009). When over-reliance on duration does not mean perceiving duration differences. In Proceedings of the Phonetics Teaching and Learning Conference (PTLC2009). http://www.phon.ucl.ac.uk/ptlc/ptlc2009/proceedings-index.html. Accessed April 04, 2018. Chambers, L., Galaczi, E., & Gilbert, S. (2012). Test-taker familiarity in speaking tests: Does it make a difference? Cambridge ESOL Research Notes forthcoming, 49, 33–40. Chamot, A. U. (2004). Issues in language learning strategy research and teaching. Electronic Journal of Foreign Language Teaching, 1(1), 14–26. Chan, A. Y. W. (2004). Syntactic transfer: Evidence from the interlanguage of Hong Kong Chinese ESL learners. Modern Language Journal, 88, 56–74. https://doi.org/10.1111/j.00267902.2004.00218.x. Chapetón, C. M. (2008). Cross-linguistic influence in the writing of an Italian learner of English as a foreign language: An exploratory study. Colombian Applied Linguistics Journal, 10, 52–70. Chládková, K., & Podlipský, V. J. (2011). Native dialect matters: Perceptual assimilation of Dutch vowels by Czech listeners. The Journal of the Acoustical Society of America, 130(4), EL186– EL192. Coe, N. (2001). Speakers of Spanish and Catalan. In M. Swan & B. Smith (Eds.), Learner English A teacherʼs guide to interference and other problems (2nd ed.). Cambridge: Cambridge University Press. Colantoni, L., Steele, J., & Escudero, P. (2015). Second language speech. Cambridge: Cambridge University Press. Commission of the European Communities. (2006). Debating Europe, involving people. On European Communication Policy (White Paper 35). Retrived from http://europa.eu/documents/ comm/white_papers/pdf/com2006_35_en.pdf. Cook, V. (2001). Using the first language in the classroom. Canadian Modern Language Review, 57(3), 402–423. Cook, V. J. (1991). The poverty-of-the-stimulus argument and multicompetence. Second Language Research, 7, 103–117. Cook, V. J. (1992). Evidence for multicompetence. Language Learning, 42, 557–591. Cook, V. J. (2003). Effects of the second language on the first. Clevedon: Multilingual Matters. Costa, A., Colomé, A., & Caramazza, A. (2000). Lexical access in speech production: The bilingual case. Psicológica, 21, 403–437. Council of Europe. (2001). Common European framework of reference for languages: Learning, teaching, assessment. Cambridge University Press. Coyle, D. (2007). Content and language integrated learning: toward a connected research agenda for CLIL pedagogies. The International Journal of Bilingual Education and Bilingualism, 10, 543–562. Coyle, D., Hood, P., & Marsh, D. (2010). CLIL: Content and Language Integrated Learning. Cambridge: Cambridge University Press. Crossley, S. A., & McNamara, D. S. (2012). Detecting the first language of second language writers using automated indices of cohesion, lexical sophistication, syntactic complexity and conceptual knowledge. In S. Jarvis & S. A. Crossley (Eds.), Approaching language transfer through text classification: Explorations in the detection-based approach (pp. 106–126). Bristol, UK: Multilingual Matters. Cummins, J. (1978). Bilingualism and the development of meatlisntuistic awareness. Journal of Cross-Cultural Psychology, 9, 131–149.

References

239

Cummins, J. (2007). Rethinking monolingual instructional strategies in multilingual classrooms. Canadian Journal of Applied Linguistics/Revue Canadienne de Linguistique Appliquée, 10(2), 221–240. Cutillas, J. A. (2001). La adquisición de onsets complejos en inglés por hablantes de español: Sonoridad y optimalidad. In M. L Pascual, P. Guerrero, M. C. Portero, & A. Ruiz-Sánchez (Eds.), Estudios de Filología Inglesa. Actas de las IV Jornadas de Filología Inglesa (pp. 105– 120). Córdoba: Servicio de Publicaciones. Universidad de Córdoba. Cutler, A., & Pasveer, D. (2006). Explaining cross-linguistic differences in effects of lexical stress on spoken-word recognition. In R. Hoffman & H. Mixdorff (Eds.), Proceedins of the 3rd international conference on speech prosody (pp. 237–400). Dresden: TUD press. Dalton-Puffer, C. (2011). Content-and-language integrated learning: From practice to principles? Annual Review of Applied Linguistics, 31, 182–204. Dastjerdi, H. V., & Nasri, N. (2013). A cross-cultural study of speech acts: Situational aspects of congratulations in English, Persian and Arabic. Global Journal of Foreign Language Teaching, 1, 15–22. De Angelis, G., & Selinker, L. (2001). Interlanguage transfer and competing linguistic systems in the multilingual mind. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Cross-linguistic influence in third language acquisition. Psycholinguistic perspectives (pp. 42–58). Clevedon: Multilingual Matters. De Angelis, G. (2005). Multilingualism and non-native lexical transfer: An identification problem. International Journal of Multilingualism, 2, 1–25. https://doi.org/10.1080/ 17501220508668374. De Angelis, G. (2007). Third or additional language acquisition. Clevedon: Multilingual Matters. De Bot, K., & Jaensch, C. (2015). What is special about L3 processing? Bilingualism: Language and Cognition, 18, 130–144. De Pablos Ortega, C. (2011). Análisis sociopragmático del habla expresivo de agradecimiento en español. Actas del III Coloquio Internacional del Programa EDICE (pp. 685–691). Valencia: Universidad de Valencia. Derwing, T. M., & Munro, M. J. (2005). Second language accent and pronunciation teaching: A research-based approach. Tesol Quarterly, 39(3), 379–397. Devos, N. J. (2016). Peer interactions in new content and language integrated settings. Biefeld: Springer. Dewaele, J.-M. (1998). Lexical inventions: French interlanguage as L2 versus L3. Applied Linguistics, 19, 471–490. https://doi.org/10.1093/applin/19.4.471. Di Camilla, F., & Antón, M. (2012). Functions of L1 in the collaborative interaction of beginning and advanced second language learners. International Journal of Applied Linguistics, 22(2), 160–188. Dijkstra, T., Timmermans, M., & Schriefers, H. (2000). On being blinded by your other language: Effects of task demands on interlingual homograph recognition. Journal of Memory and Language, 42(4), 445–464. https://doi.org/10.1006/jmla.1999.2697. Dijkstra, T., Van Jaarsveld, H., & Brinke, S. T. (1998). Interlingual homograph recognition: Effects of task demands and language intermixing. Bilingualism: Language and Cognition, 1(1), 51–66. http://doi.org/10.1017/S1366728998000121. Döpke, S. (1998). Competing language structures: The acquisition of verb placement by bilingual German-English children. Journal of Child Language, 25(3), 555–584. Dörnyei, Z., & Shekhan, P. (2003). Individual differences in second language learning. In C. J. Doughty & M. H. Long (Eds.), The handbook of second language acquisition (pp. 589–601). Malden, MA: Blackwell. Duff, P. A., Rossiter, M. J., Derwing, T. M., & Jones, V. M. (2008). Is a picture worth a thousand words? TESOL Quarterly, 42(2), 325–329. Dulay, H., & Burt, M. (1974a). Errors and strategies in child second language acquisition. TESOL Quarterly, 8(2), 129–136.

240

References

Dulay, H., & Burt, M. (1974b). Natural sequences in child second language acquisition. Language Learning, 24, 37–53. Ecke, P. (2001). Lexical retrieval in a third language: Evidence from errors and tip-of-the-tongue states. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Cross-linguistic influence in third language acquisition: Psycholinguistic perspectives (pp. 90–114). Clevedon: Multilingual Matters. Ecke, P. (2008). Cross-linguistic influence on word search in tip-of-the-tongue states. TESOL Quarterly, 43, 515–527. Ecke, P. (2015). Parasitic vocabulary acquisition, cross-linguistic influence, and lexical retrieval in multilinguals. Bilingualism: Language and Cognition, 18(2), 145–162. Ecke, P., & Hall, C. J. (2013). Tracking tip-of-the-tongue states in a multilingual speaker: Evidence of attrition or instability in lexical systems? International Journal of Bilingualism, 17 (6), 734–751. Eckman, F. (1977). Markedness and the contrastive analysis hypothesis. Language Learning, 27, 315–330. Economidou-Kogetsidis, M. (2013). Strategies, modification and perspective in native speakers’ requests: A comparison of WDCT and naturally occurring requests. Journal of Pragmatics, 53, 21–38. Ellis, R. (2004). Individual differences in second language learning. In A. Davies & C. Elder (Eds.), The handbook of applied linguistics (pp. 525–551). Oxford: Blackwell. Elvin, J. (2016). The role of the native language in non-native perception and spoken word recognition: English versus Spanish learners of Portuguese. Ph.D. Dissertation, Western Sydney University. Elvin, J., Escudero, P., & Vasiliev, P. (2014). Spanish is better than English for discriminating Portuguese vowels: Acoustic similarity versus vowel inventory. Frontiers in Psychology, 5, 1188. https://doi.org/10.3389/fpsyg.2014.01188. Elvin, J., Escudero, P., Williams, D., & Best, C. T. (2016). The relationship between Australian English speakers’ non-native perception and production of Brazilian Portuguese vowels. In C. Carignan (Ed.), Proceedings of the 16th Australasian International Conference on Speech Science and Technology (pp. 293–296). Parramatta: Australia. Elvin, J., Williams, D., & Escudero, P. (2016). The relationship between perception and production of Brazilian Portuguese vowels in European Spanish monolinguals. Loquens, 3, 2, e031. http://dx.doi.org/10.3989/loquens.2016.031. Elwood, K. (2004). “Congratulations”. A cross-cultural analysis of responses to another’s happy news. Retrieved from http://jairo.nii.ac.jp/0069/00000372/en. Epstein, S. D., Flynn, S., & Martohardjono, G. (1996). Second language acquisition: Theoretical and experimental issues in contemporary research. Behavioral and Brain Sciences, 1, 677–758. Erdener, D., & Burnham, D. (2005). The role of audiovisual speech and orthographic information in non-native speech production. Language Learning, 55(2), 191–228. https://doi.org/10.1111/ j.0023-8333.2005.00303.x. Escudero, P, Benders, T., & Lipski, S. C. (2009). Native, non-native and L2 perceptual cue weighting for Dutch vowels: The case of Dutch, German, and Spanish listeners. Journal of Phonetics, 37(4), 452–465. http://dx.doi.org/10.1016/j.wocn.2009.07.006. Escudero, P. (2001). The role of the input in the development of L1 and L2 sound contrasts: Language-specific cue weighting for vowels. In A. H.-J. Do, L. Dominguez, & A. Johansen (Eds.), Proceedings of the 25th annual Boston University conference on language development (pp. 250–261). Somerville, MA: Cascadilla Press. Escudero, P., & Boersma, P. (2004). Bridging the gap between L2 speech perception research and phonological theory. Studies in Second Language Acquisition, 26(4), 551–585. Escudero, P. (2005). Linguistic perception and second language acquisition. Ph.D. Dissertation, Utrecht University. Escudero, P. (2006). Second language phonology: The role of perception. In M. C. Pennington (Ed.), Phonology in context (pp. 109–134). New York: Palgrave Macmillan.

References

241

Escudero, P. (2009). The linguistic perception of similar L2 sounds. Phonology in Perception, 15, 152–190. Escudero, P., & Chládková, K. (2010). Spanish listeners’ perception of American and Southern British English vowels. The Journal of the Acoustical Society of America, 128(5), EL254–9. http://doi.org/10.1121/1.3488794. Escudero, P., & Hoffmann Bion, R. A. (2007). Modelling vowel normalization and sound perception as sequential processes. In J. Trouvain, & W. J. Barry (Eds.), Proceedings of the 16th international congress of phonetic sciences (pp. 1413–1416). Saarbrücken: University of Saarbrucken. Escudero, P., & Vasiliev, P. (2011). Cross-language acoustic similarity predicts perceptual assimilation of Canadian English and Canadian French vowels. The Journal of the Acoustical Society of America, 130(5), EL277–EL283. Escudero, P., & Williams, D. (2011). Perceptual assimilation of Dutch vowels by Peruvian Spanish listeners. The Journal of the Acoustical Society of America, 129(1), EL1–EL7. Escudero, P., & Williams, D. (2012). Native dialect influences second-language vowel perception: Peruvian versus Iberian Spanish learners of Dutch. The Journal of the Acoustical Society of America, 131(5), EL406–12. http://doi.org/10.1121/1.3701708. Escudero, P., & Williams, D. (2014). Distributional learning has immediate and long-lasting effects. Cognition, 13(2), 408–413. https://doi.org/10.1016/j.cognition.2014.07.002. Escudero, P., Benders, T., & Wanrooij, K. (2011). Enhanced bimodal distributions facilitate the learning of second language vowels. The Journal of the Acoustical Society of America, 130(4), EL206–12. http://doi.org/10.1121/1.3629144. Escudero, P., Broersma, M., & Simon, E. (2013). Learning words in a third language: Effects of vowel inventory and language proficiency. Language and Cognitive Processes, 28(6), 746–761. https://doi.org/10.1080/01690965.2012.662279. Escudero, P., Ellen, S. & Mitterer, H. (2012). The perception of English front vowels by North Holland and Flemish listeners: Acoustic similarity predicts and explains cross-linguistic and L2 perception. Journal of Phonetics, 40(2), 280–288. http://doi.org/http://dx.doi.org/10.1016/j. wocn.2011.11.004. Escudero, P., Sisinni, B., & Grimaldi, M. (2014). The effect of vowel inventory and acoustic properties in Salento Italian learners of Southern British English vowels. The Journal of the Acoustical Society of America, 135(3), 1577–1584. https://doi.org/10.1121/1.4864477. Eslami-Rasekh, A., & Mardani, M. (2010). Investigating the effects of teaching apology speech act, with a focus on intensifying strategies, on pragmatic development of EFL learners: The Iranian Context. International Journal of Language Society and Culture, 30, 96–103. Estebas, E. (2012). Teach yourself English pronunciation: An interactive course for Spanish speakers (2nd ed.). Oleiros: Netbiblo. Ethnologue: Languages of the World https://www.ethnologue.com/statistics/family. Faerch, C., & Kasper, G. (1983). Strategies in interlanguage communication. London: Longman. Falk, Y. (2010). Gingerly studied transfer phenomena in L3 German syntax: The role of the second language in third language acquisition. Ph.D. dissertation. Radbout Universiteit Nijmegen. LOT. Falk, Y., & Bardel, C. (2010). The study of the role of the background languages in third language acquisition. The state of the art. International Review of Applied Linguistics and Teaching, 48, 185–219. Falk, Y., & Bardel, C. (2011). Object pronouns in German L3 syntax: Evidence for the L2 status factor. Second Language Research, 27, 59–82. https://doi.org/10.1177/0267658310386647. Falk, Y., Lindqvist, C., & Bardel, C. (2015). The role of L1 explicit metalinguistic knowledge in L3 oral production at the initial state. Bilingualism: Language and cognition, 18, 227–235. Fallah, N., Jabbari, A., & Fazilatfar, A. M. (2016). Source(s) of syntactic CLI: The case of L3 acquisition of English possessives by Mazandarani-Persian bilinguals. Second Language Research, 32, 225–245.

242

References

Faucette, P. (2001). A pedagogical perspective on communication strategies: Benefits of training and an analysis of English language teaching materials. Second Language Studies, 19(2), 1–40. Félix-Brasdefer, J. C. (2007). Pragmatic development in the Spanish as an EFL classroom: A cross-sectional study of learner requests. Intercultural Pragmatics, 4(2), 253–286. Fernández Dobao, A. M. (2002). The effect of language proficiency on communication strategy use: A case study of Galician learners of English. Miscelánea: A Journal of English and American Studies, 25, 53–75. Flege, J. E. (1995). Second language speech learning: Theory, findings, and problems. In W. Strange (Ed.), Perception and linguistic experience: Issues in cross-language research (pp. 233–276). Timonium, MD: York Press. Flege, J. E. (1997). English vowel production by Dutch talkers: more evidence for the “new” versus “similar” distinction. In A. James, & J. Leather (Eds.), Second-language speech, structure and process (pp. 11–52). Berlin: Mouton de Gruyter. Flege, J. E. (2007). Language contact in bilingualism: Phonetic system interactions. In J. Cole & J. I. Hualde (Eds.), Laboratory phonology 9 (pp. 353–381). New York: Mouton de Gruyer. Flynn, S. (1983). A study of the effects of principal branching direction in second language acquisition: The generalization of a parameter of Universal Grammar from first to second language acquisition. Ithaca, NY: Cornell University dissertation. Flynn, S. (1987). A parameter-setting model of L2 acquisition: Experimental studies in anaphora. Dordrecht: Reidel Press. Flynn, S. (1988). The nature of development in L2 acquisition and implications for theories of language acquisition in general. In S. Flynn & W. O’Neil (Eds.), Linguistic theory in second language acquisition (pp. 76–90). The Netherlands: Kluwer Academic Publishers. Flynn, S. (1989). Spanish, Japanese and Chinese speakers’ acquisition of English relative clauses: New evidence for the head-direction parameter. In K. Hyltenstam & L. K. Obler (Eds.), Bilingualism across the lifespan: Aspects of acquisition, maturity, and loss (pp. 116–131). Cambridge: Cambridge University Press. Flynn, S. (2018). The more languages you know, the easier it gets. In Plenary presentation CILEX Conference, Monterrey, Mexico November 10, 2018. Flynn, S., & Berkes, E. (2017). Toward a new understanding of syntactic CLI. Evidence from L2 and L3 acquisition. In T. Angelovska & A. Hahn (Eds.), L3 syntactic transfer: Models, new developments and implications (pp. 35–61). Amsterdam: John Benjamins Publishing Company. Flynn, S., & O’Neil, W. (Eds.). (1988). Linguistic theory in second language acquisition. The Netherlands: Kluwer Academic Publishers. Flynn, S., Foley, C., & Vinnitskaya, I. (2004). The cumulative-enhancement model of language acquisition: Comparing adults’ and children’s patterns of development in first, second and third language acquisition of relative clauses. International Journal of Multilingualism, 1, 3–16. https://doi.org/10.1080/14790710408668175. Foote, R. (2009). Transfer and L3 acquisition: The role of typology. In Y.-K. I. Leung (Ed.), Third language acquisition and Universal Grammar (pp. 89–114). Clevedon: Multilingual Matters. Forsyth, H. (2014). The Influence of L2 transfer on L3 English written production in a bilingual German/Italian population: A study of syntactic errors. Open Journal of Modern Linguistics, 4, 429–456 Fries, C. (1945). Teaching and learning English as a foreign language. Ann Arbor, MI: University of Michigan Press. Gabrys-Barker, D. (2006). The interaction of languages in the lexical search of multilingual language users. In J. Arabski (Ed.), Cross-linguistic influences in the second language lexicon (pp. 144–166). Clevedon: Multilingual Matters. Galambos, S. J., & Hakuta, K. (1988). Subject-specific and task. Specific characterisitcs of meatalinguistic awareness in bilingual children. Applied Psycholinguistics, 9, 141–162.

References

243

Gallardo-del-Puerto, F. (2015). L1 influence in CLIL vs. EFL schoolchildren: A study of codeswitching and transfer lapses. Paper presented at the 33rd AESLA International Conference, Technical University of Madrid, 16–18 April. Gallardo-del-Puerto, F., & Friedman, K. (2013). On the pronunciation of L2 English word-final consonant clusters in monomorphemic versus inflected words. In R. Arias, M. LópezRodríguez, A. Moreno, & C. Pérez-Hernández (Eds.), Hopes and fears: English and American studies in Spain (pp. 226–231). Málaga: Departamento de Filología Inglesa, Francesa y Alemana, Universidad de Málaga. Gallardo-del-Puerto, F., García-Lecumberri, M. L, & Cenoz, J. (2005). Degree of foreign accent and age of onset in formal school instruction. Proceedings of the Phonetics Teaching and Learning Conference (PTLC2005).http://www.phon.ucl.ac.uk/home/johnm/ptlc2005/pdf/ ptlcp48.pdf. Accessed April 04, 2018. Ganuza, N. (2008). Syntactic variation in the Swedish of adolescents in multilingual urban settings. Subject-verb order in declaratives, questions and subordinate clauses. Doctoral dissertation. Stockholm: Stockholm University, Center for Research on Bilingualism. García Mayo, M. P. (2015). The interface between task-based language teaching and content-based instruction—Introduction to Special Issue on learning language and content through tasks: Exploring the interfaces. System, 54, 1–3. García Mayo, M. P., & Hidalgo, M. A. (2017). L1 use among young EFL mainstream and CLIL learners in task-supported interaction. System, 67, 132–145. García Mayo, M. P., & Imaz Agirre, A. (2017). Child EFL interaction; age, instructional setting and development. In J. Enever & E. Lindgren (Eds.), Early language learning: Complexity and mixed methods (pp. 249–268). Bristol: Multilingual Matters. García Mayo, M. P., & Lázaro-Ibarrola, A. (2015). Do children negotiate for meaning in task-based interaction? Evidence from CLIL and EFL settings. System, 54, 40–54. http://doi. org/10.1016/j.system.2014.12.001. García Mayo, M. P. (2018). Child task-based interaction in EFL settings: Research and challenges. International Journal of English Studies, 18(2), 119–143. García Mayo, M. P., & González Alonso, J. (2015). Introduction: Third language (L3) acquisition: A focus on cognitive approaches. Bilingualism: Language and Cognition, 18, 127–129. García Mayo, M.P., & Rothman, J. (2012). L3 morphosyntax in the generative tradition: The initial stages and beyond. In J. Cabrelli Amaro, S. Flynn, & J. Rothman (Eds.), Third language acquisition in adulthood (pp. 9–32). Amsterdam: John Benjamins. García, C. (2009). ¿Qué? ¿Cómo que te vas a casar? Congratulations and rapport management: A case study of Peruvian Spanish speakers. Pragmatics, 19(2), 197–222. García-Landa, L. (1999). Efecto de la enseñanza sistemática en la percepción y pronunciación en inglés para hispanohablantes: Los alomorfos de los pasados regulares. Estudios de Lingüística Aplicada, 30(31), 63–90. https://doi.org/10.22201/enallt.01852647p.2009.50.536. Gass, S. (1984). A review of interlanguage syntax: language transfer and language universals. Language Learning, 34(2), 115–132. https://doi.org/10.1111/j.1467-1770.1984.tb01007.x. Gass, S. (1988). Second language acquisition and linguistic theory: The role of languagetransfer. In S. Flynn & W. O’Neil (Eds.), Linguistic theory in second language acquisition (pp. 384– 403). The Netherlands: Kluwer Academic Publishers. Gass, S., & Varonis, E. (1985). Task variation and nonnative/nonnative negotiation of meaning. In S. M. Gass & C. G. Madden (Eds.), Input in second language acquisition (pp. 149–161). Rowley, MA: Newbury House. Gené Gil, M., Garau, M. J., & Salazar Noguera, J. (2012). A case study exploring oral language choice between the target language and the L1s in mainstream CLIL and EFL secondary education. Revista de Lingüística y Lenguas Aplicadas, 7(1), 133–146. Gené Gil, M., Juan Garau, M., & Salazar Noguera, J. (2012). A case study exploring oral language choice between the target language and the L1s in mainstream CLIL and EFL secondary education. Revista de Lingüística y Lenguas Aplicadas, 7, 133–145.

244

References

Geva, E., & Shanahan, T. (2006). Second-language oral proficiency and second-language literacy. In D. August & T. Shahanan (Eds.), Developing literacy in second-language learners: Report of the national literacy panel on language-minority children and youth (pp. 123–139). Mahwah, NJ: Lawrence Erlbaum Associates. Giancaspro, D., Halloran, B., & Iverson, M. (2015a). Examining L3 transfer: The acquisition of differential object marking in L3 Brazilian Portuguese. Bilingualism: Language and Cognition, 18, 191–207. Giancaspro, D., Halloran, B., & Iverson, M. (2015b). Transfer at the initial stages of L3 Brazilian Portuguese: A look at three groups of English/Spanish bilinguals. Bilingualism: Language and Cognition, 12, 191–207. Gil, K.-H., Mardsen, H., & Whong, M. (2017). The meaning of negation in the second language classroom: Evidence from ‘any’. Language Teaching Research. https://doi.org/10.1177/ 1362168817740144. Gilichinskaya, Y. D, & Strange, W. (2010). Perceptual assimilation of American English vowels by in experienced Russian listeners. The Journal of the Acoustical Society of America, 128(2), EL80–5. http://doi.org/10.1121/1.3462988. Golden, A., Jarvis, S., & Tenfjord, K. (2017). Crosslinguistic influence and distinctive patterns of language learning findings and insights from a learner corpus. Bristol: Multilingual Matters. Gómez-Lacabex, E., & García-Lecumberri, M. L. (2007). Perception of the contrast full vowel/schwa in English by trained Spanish learners.In Proceedings of the Phonetics Teaching and Learning Conference (PTLC2007).http://www.phon.ucl.ac.uk/ptlc/ptlc2007_web_ procindex.html. Accessed April 04, 2018. Gómez-Lacabex, E., García-Lecumberri, M. L, & Cooke M. (2005). English vowel reduction by untrained Spanish learners: perception and production. In Proceedings of the Phonetics Teaching and Learning Conference (PTLC2005).https://www.phon.ucl.ac.uk/home/johnm/ ptlc2005/pdf/ptlcp45.pdf. Accessed April 04, 2018. Gonzalez Alonso, J., & Rothman, J. (2017). Coming of age in L3 initial stages transfer models: Deriving developmental predictions and looking towards the future. International Journal of Bilingualism, 21(6), 683–697. Gost, C., & Celaya, M. L. (2005). Age and the use of the L1 in EFL oral production. In M. L. Carrió Pastor (Ed.), Perspectivas Interdisciplinares de la Lingüística Aplicada (Vol, 1, pp. 129–136). Valencia: Universitat Politècnica de València. Granger, S., Dagneaux, E., Meunier, F., & Paquot, M. (2009). The international corpus of learner English. Handbook and CD-ROM. Version 2. Louvain-la-Neuve: Presses universitaires de Louvain. Green, D. W. (1998). Mental control of the bilingual lexico-semantic system. Bilingualism: Language and Cognition, 1(2), 67–81. Greenberg, J. H. (Ed.). (1966). Universals of language. Cambridge: MIT Press. Guijarro-Fuentes, P., Schmitz, K., & Müller, N. (2015). The acquisition of French in multilingual contexts. Clevedon, UK: Multilingual Matters. Gutierrez-Mangado, M. J., & Martínez-Adrián, M. (2018). CLIL at the linguistic interfaces. Journal of Immersion and Content-Based Language Education, 6(1), 85–112. Hall, C. J. (2002). The automatic cognate form assumption: Evidence for the parasitic model of vocabulary development. International Review of Applied Linguistics, 40, 69–87. Hall, C. J., & Ecke, P. (2003). Parasitism as a default mechanism in L3 vocabulary acquisition. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), The multilingual lexicon (pp. 71–85). Dordrecht: Kluwer Academic Press. Hall, C. J., Newbrand, D., Ecke, P., Sperr, U., Marchand, V., & Hayes, L. (2009). Learners’ implicit assumptions about syntactic frames in new L3 words: The role of cognates, typological proximity and L2 status. Language Learning, 59, 153–202. Hammarberg, B. (2001). Roles of 1 and L2 in L3 production and acquisition. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Cross-linguistic influence in third language acquisition: Psycholinguistic perspectives (pp. 21–41). Clevedon: Multilingual Matters.

References

245

Hammarberg, B. (2009). Processes in third language acquisition. Edinburgh: Edinburgh University Press. Hammarberg, B. (2010). The languages of the multilingual: Some conceptual and terminological issues. International Review of Applied Linguistics in Language Teaching, 48, 91–104. Hammarberg, B. (2016). Flerspråkighet och tredjespråksinlärning. Några grundbegrepp [Multilingualism and third language learning. Some fundamental notions]. In C. Bardel, Y. Falk, & C. Lindqvist (Eds.), Tredjespråksinlärning [Third language learning] (pp.33–58). Lund: Studentlitteratur. Hammarberg, B. (2018). L3, the tertiary language. In A. Bonnet & P. Siemund (Eds.), Foreign language education in multilingual classrooms (pp. 127–150). Amsterdam: John Benjamins. Harris, A. C. (2000). Word order harmonies and word order change in Georgian. In R. Sornicola, E. Poppe, & A. Haley (Eds.), Stability, variation, and change of word-order patterns over time (pp. 133–163). Amsterdam: John Benjamins. https://doi.org/10.1075/cilt.213.13har. Hawkins, R., & Chan, C. Y.-H. (1997). The partial availability of Universal Grammar in second language acquisition: The ‘Failed Functional Features Hypothesis’. Second Language Research, 13, 187–226. Hayes-Harb, R., Nicol, J., & Barker, J. (2010). Learning the phonological forms of new words: Effects of orthographic and auditory input. Language and Speech, 53(3), 367–381. https://doi. org/10.1177/0023830910371460. Heras, A., & Lasagabaster, D. (2015). The impact of CLIL on affective factors and vocabulary learning. Language Teaching Research, 19, 70–88. Herdina, P., & Jessner, U. (2002). A dynamic model of multilingualism. Perspectives in change in psycholinguistics. Clevedon: Multilingual Matters. Hermas, A. (2014). Multilingual transfer: L1 morphosyntax in L3 English. International Journal of Language Studies, 8(2), 1–24. Herwig, A. (2001). Plurilingual lexical organisation: Evidence from lexical processing in L1-L2-L3-L4 translation. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Cross-linguistic influence in third language acquisition: Psycholinguistic perspectives (pp. 115–137). Clevedon: Multilingual Matters. Hickey, L. (2005). Politeness in Spain: Thanks, but no ‘thanks’. In L. Hickey & M. Steward (Eds.), Politeness in Europe (Vol. 127, pp. 317–330). Clevedon: Multilingual Matters. Honikman, B. (1964). Articulatory settings. In D. Abercrombie, D. B. Fry, P. A. D. MacCarthy, N. C. Scott, & J. L. M. Trim (Eds.), Honour of Daniel Jones: Papers contributed on the occasion of his eightieth birthday (pp. 73–84). London: Longmans, Green and Co. Hüttner, J., & Rieder-Bünemann, A. (2010). A cross-sectional analysis of oral narratives by children with CLIL and non-CLIL instruction. In C. Dalton-Puffer, T. Nikula, & U. Smit (Eds.), Language use and language learning in CLIL classrooms (pp. 61–80). Philadelphia: John Benjamins B.V. Hyde, J. (1982). The identification of communication strategies in the interlanguage of Spanish learners of English. Anglo-American Studies, 2(1), 13–30. IBM Corp. (2010). IBM SPSS statistics for windows, Version 19.0. Armonk, NY: IBM Corp. Iglesias Diéguez, K., & Martínez-Adrián, M. (2017). The influence of CLIL on receptive vocabulary: A preliminary study. Journal of English Studies, 15, 107–134. Ionin, T., Montrul, S., & Santos, H. (2011). An experimental investigation of the expression of genericity in English, Spanish and Brazilian Portuguese. Lingua, 121, 963–985. Ioup, G. (1984). Is there a structural foreign accent? A comparison of syntactic and phonological errors in second language acquisition. Language Learning, 34, 1–17. Iverson, P., & Evans, B. G. (2009). Learning English vowels with different first-language vowel systems II: Auditory training for native Spanish and German speakers. The Journal of the Acoustical Society of America, 126(2), 866–877. https://doi.org/10.1121/1.3148196. Iverson, P., P. Kuhl, K., Akahane-Yamada, R., Diesch, E., Tohkura, Y., Kettermann, A., & Siebert, C. (2003). A perceptual interference account of acquisition difficulties for non-native phonemes. Cognition, 87, 47–57. http://doi.org/10.1016/S0.

246

References

James, A. (1996). A cross-linguistic approach to language awareness. Language Awareness, 5, 3 and 4, 138–148. James, C. (2013). Errors in language learning and use: Exploring error analysis. London: Routledge. Jarvis, S. (2000). Methodological rigor in the study of transfer: Identifying L1 influence in the interlanguage lexicon. Language Learning, 50, 245–309. Jarvis, S. (2002). Short texts, best-fitting curves, and new measures of lexical diversity. Language Testing, 19, 57–84. Jarvis, S. (2009). Lexical transfer. In A. Pavlenko (Ed.), Bilingual mental lexicon: Interdisciplinary approaches (pp. 99–124). Clevedon, UK: Multilingual Matters. Jarvis, S. (2012). The detection-based approach: An overview. In S. Jarvis & S. A. Crossley (Eds.), Approaching language transfer through text classification: Explorations in the detection-based approach (pp. 1–33). Bristol, UK: Multilingual Matters. Jarvis, S. (2016). The scope of transfer research. In L. Yu & T. Odlin (Eds.), New perspectives on transfer in second language learning (pp. 17–47). Bristol: Multilingual Matters. Jarvis, S. (2017). Transfer: An overview with an expanded scope. In A. Golden, S. Jarvis, & K. Tenfjord (Eds.), Crosslinguistic influence and distinctive patterns of language learning findings and insights from a learner corpus (pp. 12–28). Bristol: Multilingual Matters. Jarvis, S., & Crossley, S. A. (Eds.). (2012). Approaching language transfer through text classification: Explorations in the detection-based approach. Bristol, UK: Multilingual Matters. Jarvis, S., & Paquot, M. (2015). Native language identification. In S. Granger, G. Gilquin, & F. Meunier (Eds.), The Cambridge handbook of learner corpus research (pp. 605–627). Cambridge: Cambridge University Press. Jarvis, S., & Pavlenko, A. (2008). Crosslinguistic influence in language and cognition. New York: Routledge. Jarvis, S., Bestgen, Y., & Pepper, S. (2013). Maximizing classification accuracy in native language identification. In Proceedings of the Eighth Workshop on Innovative Use of NLP for Building Educational Applications (pp. 111–118). Atlanta: Association for Computational Linguistics. Jarvis, S., Castañeda-Jiménez, G., & Nielsen, R. (2012). Detecting L2 writers’ L1s on the basis of their lexical styles. In S. Jarvis & S. A. Crossley (Eds.), Approaching language transfer through text classification: Explorations in the detection-based approach (pp. 34–70). Bristol, UK: Multilingual Matters. Jessner, U. (2006). Linguistic awareness in multilinguals. Edinburgh: Edinburgh University Press. Jessner, U. (2008). Teaching third languages: Findings, trends and challenges. Language Teaching, 41(1), 15–56. https://doi.org.ezproxy.ub.gu.se/10.1017/S0261444807004739. Jessner, U., Allgaüer-Hackl, E., & Hofer, B. (2016). Emerging multilingual awareness in educational contexts: From theory to practice. The Canadian Modern Language Review/La revue canadienne des langues vivantes, 72(2), 157–182. Jiang, N. (2000). Lexical development and representation in a second language. Applied Linguistics, 21, 47–77. Jiménez Catalán, R. M. (2003). Sex differences in L2 vocabulary learning strategies. International Journal of Applied Linguistics, 13(1), 54–77. Jones, H. (2010). First language communication in the second language classroom: A valuable or damaging resource? M.A. Dissertation, Memorial University of Newfoundland, St. John’s, Canada. Jourdain, S. (2000). A native-like ability to circumlocute. The Modern Language Journal, 84(2), 185–195. Kallestinova, E. (2007). Aspects of word order in Russian. Ph.D. Dissertation. USA: University of Iowa. Kasper, G., & Dahl, M. (1991). Research methods in interlanguage pragmatics. Studies in Second Language Acquisition, 12, 215–247.

References

247

Kasper, G., & Rose, K. R. (2002). Pragmatic development in a second language. Oxford: Blackwell. Kasper, G., & Schmidt, R. (1996). Developmental issues in interlanguage pragmatics. Studies in Second Language Acquisition, 18, 149–169. Kaushanskaya, M., & Marian, V. (2007). Age-of-acquisition effects in the development of a bilingual advantage for word learning. In Proceedings of the Boston University Conference on Language Development. Kaushanskaya, M., & Marian, V. (2009). Bilingualism reduces native-language Interference during novel-word learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35, 829–835. Kecskes, I., & Papp, T. (2000). Foreign language and mother tongue. Mahwah, NJ: Erlbaum. Keidel, A., & Hörberg, T. (2017). Qualitative differences in L3 learners’ neurophysiological response to L1 versus L2 transfer. Conference paper presented at Interspeech, Stockholm 2017. Kellerman, E. (1977). Towards a characterization of the strategy of transfer in second language learning. Interlanguage Studies Bulletin, 2(1), 58–145. Kellerman, E. (1983). Now you see it, now you don’t. In S. M. Gass & L. Selinker (Eds.), Language transfer in language learning (pp. 112–134). Rowley, MA: Newbury House. Kemp, C. (2001). Metalinguistic awareness in multilinguals: Implicit and explicit grammatical awareness and its relationship with language experience and language attainment. Ph.D. dissertation. Unviersity of Edinburgh. Kenworthy, J. (1987). Teaching english pronunciation. New York: Longman Group UK Limited. Khanji, R. (1993). Interlanguage talk: The Relation between task types and CS among EFL Arab learners. In J. E. Alatis (Ed.), Georgetown University round table on language and linguistics 1993 (pp. 428–436). Washington D.C.: Georgetown U. P. Kivistö-de Souza, H., & Mora, J.C. (2014). Effects of speech rate on the intelligibility of non-native vowels. Concordia Working Papers in Applied Linguistics, 5, 337–352. Koike, D. A., & Palmiere, D. T. L. (2011). First and second language pragmatics in third language oral and written modalities. Foreign Language Annals, 44(1), 80–104. Kroll, J. F., Bogulski, C. A., & McClain, R. (2012). Psycholinguistic perspectives on second language learning and bilingualism: The course and consequence of cross-language competition. Linguistic Approaches to Bilingualism, 2, 1–24. Kroll, J., & Dussias, P. (2017). The benefits of multilingualism to the personal and professional development of residents of the US. Foreign Language Annals, 50, 248–259. Kulundary, V., & Grabriele, A. (2012). Examining the role of syntactic development in the L2 on the acquisition of an L3: A look at relative clauses. In J. Cabrelli-Amaro, S. Flynn, & J. Rothman (Eds.), Third language acquisition in adulthood (pp. 195–222). Amsterdam: John Benjamins. Lado, R. (1957). Linguistics across cultures. Ann Arbor, MI: University of Michigan Press. Lammiman, K. (2010). Cross linguistic influence of an L3 on L1 and L2. Innervate Leading Undergraduate Work in English Studies, 2, 274–283. Lantolf, J. P., & Thorne, S. L. (2007). Sociocultural theory and second language learning. In B. van Patten & J. Williams (Eds.), Theories in second language acquisition (pp. 201–224). Mahwah, NJ: Lawrence Erlbaum. Lara, R., Mora, J. C., & Pérez-Vidal, C. (2014). How long is long enough? L2 English development through study abroad programmes varying in duration. Innovation in Language Learning and Teaching, 9(1), 46–57. https://doi.org/10.1080/17501229.2014.995764. Larson-Hall, J. (2010). A guide to doing statistics in second language research using SPSS. New York: Routledge. Lasagabaster, D. (1997). Creatividad y conciencia metalingüística: incidencia en el aprendizaje del inglés como L3. Ph.D. dissertation. Leioa. Universidad del País Vasco. Lasagabaster, D. (2007). Trilingüismo escolar: sobre la presencia del inglés en el sistema educativo vasco. Hizkunea, 10, 1–13.

248

References

Lasagabaster, D. (2008). Foreign language competence in content and language integrated courses. The Open Applied Linguistics Journal, 1, 31–42. Lasagabaster, D. (2013). The use of the L1 in CLIL classes: The teachers’ perspective. LACLIL, Latin American Journal of Content and Language Integrated Learning, 6, 1–21. Lázaro Ibarrola, A. (2016). Are CLIL learners simply faster or also different? Evidence from L1 use in the repair sequences and discourse markers of CLIL and EFL learners. Vigo International Journal of Applied Linguistics, 13, 127–145. Lázaro-Ibarrola, A., & Azpilicueta-Martínez, R. (2015). Investigating negotiation of meaning in EFL children with very low levels of proficiency. International Journal of English Studies, 15(1), 1–21. Lázaro-Ibarrola, A., & García Mayo, M. P. (2012). L1 use and morphosyntactic development in the oral production of EFL learners in a CLIL context. International Review of Applied Linguistics in Language Teaching, 50(2), 135–160. Lázaro-Ibarrola, A., & Hidalgo, M. A. (2017). Benefits and limitations of conversational interactions among young learners of English in a CLIL context. In M. P. García Mayo (Ed.), Learning foreign languages in primary school: Research insights (pp. 86–102). Bristol: Multilingual Matters. Leal, T., & Slabakova, R. (2017). The relationship between L2 instruction, exposure, and the L2 acquisition of a syntax-discourse property of L2 Spanish. Language Teaching Research. https://doi.org/10.1177/1362168817745714. Leech, G. (1983). Principles of pragmatics. London: Longman. Leung, Y.-K. (2006). Full transfer versus partial transfer in L2 and L3 acquisition. In R. Slabakova, S. Montrul, & P. Prévost (Eds.), Inquiries in linguistic development: In honor of Lydia White (pp. 157–187). Amsterdam: John Benjamins. Leung, Y-k. I. (2005). L2 versus L3 initial state: A comparative study of the acquisition of French DPs by Vietnamese monolinguals and Cantonese-English bilinguals. Bilingualism: Language and Cognition, 8, 39–61. Levelt, J., Willem, M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. Behavioural and Brain Sciences, 22, 1–75. Levis, J. M. (2016). Research into practice: How research appears in pronunciation teaching materials. Language Teaching, 49(3), 423–437. Levis, J. M., & Wu, A. (2018). Pronunciation—Research into practice and practice into research. CATESOL Journal, 30(1), 1–12. Lightbown, P. (2000). Anniversary article: Classroom SLA research and second language teaching. Applied Linguistics, 21, 431–462. Lightbown, P., & Spada, N. M. (2006). How languages are learned. USA: Oxford University Press. Lin, A. M. (2015). Conceptualising the potential role of L1 in CLIL. Language, Culture and Curriculum, 28(1), 74–89. Lindqvist, C. (2009). The use of the L1 and the L2 in French L3. Examining cross-linguistic lexemes in multilingual learners’ oral production. International Journal of Multilingualism, 6(3), 281–297. Lindqvist, C. (2010). Inter- and intralingual lexical influences in advanced learners’ French L3 oral production. Special issue of the International Review of Applied Linguistics in Language Teaching, 48(2–3), 131–157. (Approaches to third language acquisition). Lindqvist, C. (2012). Advanced learners’ word choices in French L3 oral production. In J. Cabrelli Amaro, S. Flynn, & J. Rothman (Eds.), Third language acquisition in adulthood (pp. 255– 280). Amsterdam: John Benjamins. Lindqvist, C. (2015). Do learners transfer from the language they perceive as most closely related to the L3? The role of psychotypology for lexical and grammatical cross-linguistic influence in French L3. In G. De Angelis, U. Jessner, & M. Kresic (Eds.), Multilingualism: Crosslinguistic influence in language learning (pp. 231–251). New York, London: Bloomsbury Publishing.

References

249

Lindqvist, C., & Bardel, C. (2013). Exploring the impact of the proficiency and typology factors: Two cases of multilingual learners’ L3 learning. In M. Pawlak, & L. Aronin (Eds.), Essential topics in applied linguistics and multilingualism. Studies in honor of David Singleton (pp. 253– 266). Switzerland: Springer International Publishing. Lindqvist, C., & Falk, Y. (2014). When Germans begin to learn Swedish. Which is the transfer source for function words, content words and syntax? In L. Roberts, I. Vedder, & J. Hulstijn (Eds.), Eurosla yearbook 14 (pp. 225–239). Amsterdam: John Benjamins. Lindqvist, C., & Falk, Y. (2017). Vad är ett andraspråk, egentligen? In S. Bendegard, U. Melander Marttala, & M. Westman (Eds.), Language and norms. Papers from the ASLA Symposium at Uppsala University April 21–22, 2016 (pp. 71–77) (ASLA:s skriftserie nr 26.) Uppsala 2017. Liskin Gasparro, J. E. (1996). Circumlocution, communication strategies, and the ACTFL proficiency guidelines: An analysis of student discourse. Foreign Language Annals, 29(3), 317–330. Little, D. (2016). Languages at school: A challenge for multilingual cities. In L. King & L. Carson (Eds.), The multilingual city: Vitality, conflict and change (pp. 149–176). Bristol: Multilingual Matters. Littlewood, W., & Yu, B. (2011). First language and target language in the foreign language classroom. Language Teaching, 44(1), 64–77. Llanes, À., Mora, J. C., & Serrano, R. (2017). Differential effects of SA and intensive AH courses on teenagers’ L2 pronunciation. International Journal of Applied Linguistics, 27(2), 470–490. https://doi.org/10.1111/ijal.12151. Llurda, E. (1997). Pronunciation and intelligibility of Hispanic learners of English. In J. M. Oro & J. Varela (Eds.), Adquisición y aprendizaje de lenguas segundas y sus literaturas (pp. 81–86). Santiago de Compostela: Servizo de Publicacións da Universidade de Santiago de Compostela. Lo, Y. Y., & Lin, A. M. Y. (2019). Curriculum genres and task structure as frameworks to analyse teachers’ use of L1 in CBI classrooms. International Journal of Bilingual Education and Bilingualism, 22(1), 78–90. Logan, J. S., Lively, S. E., & Pisoni, D. B. (1991). Training Japanese listeners to identify English/r/and/l: A first report. The Journal of the Acoustical Society of America, 89(2), 874–886. Long, M. H. (1996). The role of the linguistic environment in second language acquisition. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of research on language acquisition (pp. 413–468). New York: Academic Press. Lyster, R., & Izquierdo, J. (2009). Prompts versus recasts in dyadic interaction. Language Learning, 59(2), 453–498. Lyster, R. (2001). Negotiation of form, recasts, and explicit correction in relation to error types and learner repair in immersion classrooms. Language Learning, 51(1), 265–301. Lyster, R. (2007). Learning and teaching languages through content: A counterbalanced approach. Amsterdam: John Benjamins Publishing Company. Macaro, E. (2005). Codeswitching in the L2 classroom: A communication and learning strategy. In E. Llurdá (Ed.), Non-native language teachers: Perceptions, challenges, and contributions to the professions (pp. 63–84). New York: Springer. Mackey, A., & Oliver, R. (2002). Interactional feedback and children’s L2 development. System, 30(4), 459–477. Mackey, A., Oliver, R., & Leeman, J. (2003). Interactional input and the incorporation of feedback: An exploration of NS–NNS and NNS–NNS adult and child dyads. Language Learning, 53(1), 35–66. Mackey, W. F. (1967). Bilingualism as a world problem. Montreal: Harvest House. Malmasi, S., & Mark, D. (2014). Language transfer hypotheses with linear SVM weights. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) (pp. 1385–1390).

250

References

Malmasi, S., Evanini, K., Cahill, A., Tetrault, J., Pugh, R., Hamill, C., Napolitano, D., & Qian, Y. (2017). A report on the 2017 native language identification shared task. In Proceedings of the 12th Workshop on Innovative Use of NLP for Building Educational Applications (pp. 62–75). Malmasi, S., Tetreault, J., & Dras, M. (2015). Oracle and human baselines for native language identification. In Proceedings of the 10th Workshop on Innovative Use of NLP for Building Educational Applications (pp. 172–178). Manning, C. D., Raghavan, P., & Schütze, H. (2008). Introduction to information retrieval. Cambridge: Cambridge University Press. Marks, J., & Bowen, T. (2012). The book of pronunciation. Proposals for a practical pedagogy. Peaslake: Delta Publishing. Marsden, H., & Slabakova, R. (2018). Grammatical meaning and the second language classroom. Special issue of Language Teaching Research (http://journals.sagepub.com/doi/full/10.1177/ 1362168817752718). Martín Leralta, S. (2007). El desarrollo de la competencia estratégica en el aula de lengua extranjera. Un modelo de entrenamiento de estrategias de comprensión auditiva en español, Universität Bielefeld. Bielefeld, Bielefelder Server für Online-Publikationen. Martínez-Adrián, M. (2010). On L2 English transfer effects in L3 syntax. Vigo International Journal of Applied Linguistics, 7, 75–98. Martínez-Adrián, M. (2018). ¿Los juntamos? A study of L1 use in interactional strategies in CLIL vs. NON-CLIL primary school learners. International Review of Applied Linguistics in Language Teaching IRAL.http://doi.org/10.1515/iral-2015-0120. Martínez-Adrián, M. (in press). ¿Los juntamos? A study of L1 use in interactional strategies in CLIL versus non-CLIL primary school learners. International Review of Applied Linguistics in Language Teaching (IRAL). Martínez-Adrián, M., & Gallardo-del-Puerto, F. (2017). The effects of language typology on L2 lexical availability and spelling accuracy. International Journal of English Studies, 17(2), 63–79. Martínez-Adrián, M., Gallardo-del-Puerto, F., & Basterrechea, M. (2017). On self-reported use of communication strategies by CLIL learners in primary education. Language Teaching Research. https://doi.org/10.1177/1362168817722054. Martínez-Adrián, M., Gallardo-del-Puerto, F., & Basterrechea, M. (2019). On self-reported use of communication strategies by CLIL learners in primary education. Language Teaching Research, 23(1), 39–57. Martínez-Adrián, M., Gallardo-del-Puerto, F., & Gutierrez-Mangado, M. J. (2013). Phonetic and syntactc transfer effects in the English interlanguage of Basque/Spanish bilinguals. Vigo International Journal of Applied Linguistics, 10, 51–84. Martínez-Adrián, M., & Gutiérrez-Mangado, M. J. (2015). L1 use, lexical richness, accuracy and complexity in CLIL and non-CLIL learners. Atlantis, Journal of the Spanish Association for Anglo-American Studies, 37(2), 177–199. Maye, J., Werker, J. F., & Gerken, L. (2002). Infant sensitivity to distributional information can affect phonetic discrimination. Cognition, 82(3), b101–b111. Mayer, M. (1969). Frog, where are you?. New York: Dial Books for Young Readers. Mayr, R., & Escudero, P. (2010). Explaining individual variation in L2 perception: Rounded vowels in English learners of German. Bilingualism: Language and Cognition, 13(3), 279– 297. http://doi.org/10.1017/S1366728909990022. Mazurkewich, I. (1988). The acquisition of infinitive and gerund complements by second language learners. In S. Flynn & W. O’Neil (Eds.), Linguistic theory in second language acquisition (pp. 127–144). The Netherlands: Kluwer Academic Publishers. McGinnis, M. (1997). Case and locality in L-syntax: Evidence from Georgian. In H. Harley (Ed.), The UPenn/MIT roundtable on argument structure and aspect. MITWPL 32. McLaughlin, B., & Nayak, N. (1989). Processing a new language: Does knowing other languages make a difference? In H. W. Dechert & M. Raupach (Eds.), Interlingual processes (pp. 5–16). Tübingen: Gunter Narr.

References

251

McQueen, J. M. (2007). Eight questions about spoken-word recognition. In M. G. Gaskell (Ed.), The Oxford handbook of psycholinguistics (pp. 37–53). Oxford: Oxford University Press. McWhinney, B. (2000). The CHILDES project: Tools for analyzing talk (3rd ed.). Mahwah, NJ: Lawrence Erlbaum Associates. Meara, P. M. (1983). Vocabulary in a second language. London: Centre for Information on Language Teaching. Méndez, M. C., & Pavón, V. (2012). Investigating the coexistence of the mother tongue and the foreign language through teacher collaboration in CLIL contexts: Perceptions and practice of the teachers involved in the plurilingual programme in Andalusia. International Journal of Bilingual Education and Bilingualism, 15, 573–592. Meriläinen, L. (2010). Language transfer in the written English of Finnish students. Publications of the University of Eastern Finland. Dissertations in Education, Humanities and Theology. Moate, J. (2011). Reconceptualising the role of talk in CLIL. Apples-Journal of Applied Language Studies, 5(2), 17–35. Moattarian, A. (2013). Bidirectional crosslinguistic influence in language learning: Linguistic aspects and beyond. International Journal of Linguistics, 5(4), 38–49. https://doi.org/10.5296/ ijl.v5i4.3746. Mohanty, A. (1994). Bilingualism in a multilingual society: Psychosocial and pedagogical implications. Mysore: Central Institute of Indian Languages. Möhle, D. (1989). Multilingual interaction in foreign language production. In H. Dechert & W. Raupach (Eds.), Interlingual processes (pp. 179–194). Tubingen: Gunter Narr. Mompeán, J. A. (2001). A comparison between English and Spanish subjects’ typicality ratings in phoneme categories: A first report. International Journal of English Studies, 1(1), 115–156. https://doi.org/10.6018/ijes.1.1.47641. Mora, J. C., & Valls-Ferrer, M. (2012). Oral fluency, accuracy and complexity in formal instruction and study abroad learning contexts. TESOL Quarterly, 46(4), 610–641. https://doi. org/10.1002/tesq.34. Mora, J. C., Rochdi, Y., & Kivistö-de Souza, H. (2014). Mimicking accented speech as L2 phonological awareness. Language Awareness, 23, 57–75. https://doi.org/10.1080/09658416. 2013.863898. Müller, G. (1998). Incomplete category fronting: A derivational approach to remnant movement in German. Dordrecht: Kluwer. Muñoz, C. (2007a). CLIL: Some thoughts on its psycholinguistics principles. Revista Española de Lingüística Aplicada, Volumen monográfico, 1, 17–26. Muñoz, C. (2007b). Cross-linguistic influence and language switches in L4 oral production. Vigo International Journal of Applied Linguistics, 4, 73–94. Nasri, N., Dastjerdi, H. V., & Ghadiri, M. (2013). Congratulations across cultures: English versus American and Persian speakers. Procedia, Social and Behavioral Sciences, 70, 67–73. Nassaji, H. (2012). The relationship between SLA research and language pedagogy: Teachers’ perspectives. Language Teaching Research, 16(3), 337–365. Nation, I. S. P., & Meara, P. (2002). Vocabulary. In N. Schmitt (Ed.), An Introduction to applied linguistics (pp. 35–54). London: Edward Arnold. Nation, P. (1990). Teaching and learning vocabulary. New York: Newbury. Nation, R., & McLaughlin, B. (1986). Novices and experts: An information processing approach to the ‘good language learner’ problem. Applied Psycholinguistics, 7, 41–56. Navés, T., Miralpeix, I., & Celaya, M. L. (2005). Who transfers more… and what? Crosslinguistic influence in relation to school grade and language dominance in EFL. International Journal of Multilingualism, 2(2), 113–134. Neeleman, A., & Weerman, F. (1999). Flexible syntax. A theory of case and arguments. Dordrecht: Kluwer. Nemati, A. (2010). Proficiency and size of receptive vocabulary: Comparing EFL and ESL environments. International Journal of Education Research and Technology, 1(1), 46–53.

252

References

Neokleous, G. (2017). Closing the gap: Student attitudes toward first language use in monolingual EFL classrooms. TESOL Journal, 8(2), 314–341. Neuser, H. (2017). Source language of lexical transfer in multilingual learners. A mixed methods approach. Doctoral thesis. Stockholm university. Newmeyer, F., & Weinberger, S. (1988). The ontogenesis of the field of second language Research. In S. Flynn & W. O’Neil (Eds.), Linguistic theory in second language acquisition (pp. 34–46). The Netherlands: Kluwer Academic Publishers. Ó Laoire, M., & Singleton, D. (2009). The role of prior knowledge in L3 learning and use: Further evidence of psychotypological dimensions. In L. Aronin & B. Hufeisen (Eds.), The exploration of multilingualism (pp. 79–102). Amsterdam: John Benjamins. O’Malley, J. M., & Chamot, A. U. (1990). Learning strategies in second language acquisition. Cambridge: Cambridge University Press. Odlin, T. (1989). Language transfer: Cross-linguistic influence in language learning. Cambridge: Cambridge University Press. Odlin, T. (1996). On the recognition of transfer errors. Language Awareness, 5, 166–178. Odlin, T., & Jarvis, S. (2004). Same source, different outcomes: A study of Swedish influence on the acquisition of English in Finland. International Journal of Multilingualism, 1(2), 123–140. Oliver, R. (1995a). Negative feedback in child NS/NNS conversation. Studies in Second Language Acquisition, 18, 459–481. Oliver, R. (1995b). Negotiation and feedback in child second language acquisition (Unpublished doctoral thesis). The University of Western Australia, Perth. Oliver, R. (1998). Negotiation of meaning in child interactions. The relationship between conversational interaction and second language acquisition. Modern Language Journal, 82, 372–386. Oliver, R. (2000). Age differences in negotiation and feedback in classroom and pairwork. Language Learning, 50(1), 119–151. Oliver, R. (2002). The patterns of negotiation for meaning in child interactions. The Modern Language Journal, 86(1), 97–111. Oliver, R., & Mackey, A. (2003). Interactional context and feedback in child ESL classrooms. The Modern Language Journal, 87(4), 519–533. Oliveria, L. C. & Schleppegrell, M. J. (2015). Focus on grammar and meaning. Oxford: Oxford University Press. Otwinowska, A. (2016). Cognate vocabulary in language acquisition and use. Bristol: Multilingual Matters. Oxford, R. (1990). Language learning strategies: What every teacher should know. New York: Newbury House. Oxford, R. (2011). Teaching and researching language learning strategies. Harlow: Pearson. Oxford, R., & Schramm, K. (2008). Bridging the gap between psychological and sociocultural perspectives on L2 learner strategies. In A. D. Cohen & E. Macaro (Eds.), Language learner strategies (pp. 47–68). Oxford: Oxford University Press. Özçelik, Ö. (2013). Selectivity in L3 transfer: Effects of typological and linguistic similarity in the L3 Turkish of Uzbek-Russian bilinguals. Paper presented at the 36th Conference of Generative Linguistics in the Old World (GLOW 36), Lund. Palacios-Martínez, I. (2001). Improving learnerʼs pronunciation of English: Some reflections and some practical tips”. In J. M. Ruiz, P. Sheerin, & C. Estebánez (Eds.), Estudios de metodología de la lengua Inglesa (Vol. 2, pp. 17–40). Valladolid: Universidad de Valladolid. Panelli, L. (2014). ‘I'm happy for you!’: The speech act of congratulation in an EFL context (Unpublished M.A. dissertation). University of Barcelona. Paradis, M. (2004). A neurolinguistic theory of bilingualism. Amsterdam: Benjamins. Paradis, M. (2008). Language and communication disorders in multilinguals. In B. Stemmer & H.A. Whitaker (Eds.), Handbook of the neuroscience of language (pp. 341–349). Amsterdam: Elsevier.

References

253

Paradis, M. (2009). Declarative and procedural determinants of second languages. Amsterdam: Benjamins. Paredes García, F. (2012). Desarrollos teóricos y metodológicos recientes de los estudios de disponibilidad léxica. Revista Nebrija de Lingüística Aplicada, 11, 78–100. Paribakht, T. (1985). Strategic competence and language proficiency. Applied Linguistics, 6(2), 132–146. Parkinson, S. (1983). Fonética y fonología inglesas para hispanohablantes. Madrid: Editorial Empeño. Pastrana Izquierdo, A. (2010). Language functions in CLIL classrooms: Students’ oral production in different classroom activities. Views, Vienna English Working Papers (Current research on CLIL), 19(3), 72–82. Pavicic Takac, V. (2008). Vocabulary learning strategies and foreign language acquisition. Clevedon: Multilingual Matters. Pavón, V., & Ramos, M. C. (2018). Describing the use of the L1 in CLIL: An analysis of L1 communication strategies in classroom interaction. International Journal of Bilingual Education and Bilingualism.http://doi.org/10.1080/13670050.2018.1511681. Pavón, V., & Ramos, M. C. (2019). Describing the use of the L1 in CLIL: Analysing students’ L1 communication strategies in classroom interaction. International Journal of Bilingual Education and Bilingualism, 22(1), 35–48. Peláez Torres, M. (2017). Aproximación sociopragmática y pragmalingüística a la formulación de buenos deseos y felicitaciones en español. Pragmalingüística, 25, 467–489. Perrachione, T. K., Lee, J., Ha, L. Y., & Wong, P. C. M. (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. https://doi.org/10.1121/1. 3593366. Philp, J., & Tognini, R. (2009). Language acquisition in foreign language contexts and the differential benefits of interaction. IRAL-International Review of Applied Linguistics in Language Teaching, 47(3–4), 245–266. Philp, J., Oliver, R., & Mackey, A. (2008). Second language acquisition and the younger learner: Child’s play?. Amsterdam: John Benjamins Publishing. Pica, T., & Doughty, C. (1985). The role of group work in classroom second language acquisition. Studies in Second Language Acquisition, 7, 233–248. Pienemann, M., Johnston, M., & Brindley, G. (1988). Constructing an acquisition-based procedure for second language assessment. Studies in Second Language Acquisition, 10(2), 217–243. Pinter, A. (2006). Verbal evidence of task related strategies: Child versus adult interactions. System, 34(4), 615–630. Placencia, M. E., & Lower, A. (2013). Your kids are so Stinkin’ cute!:-): Complimenting behavior on Facebook among family and friends. Intercultural Pragmatics, 10(4), 617–646. Pladevall-Ballester, E., & Vraciu, A. (2017). Exploring early EFL: L1 use in oral narratives by CLIL and non-CLIL primary school learners. In M. P. García Mayo (Ed.), Learning foreign languages in primary school: Research insights (pp. 124–148). Bristol: Multilingual Matters. [Word order in Georgian]. IberoPochkhua, B. (1962). Caucasian Linguistics, 13, 109–125. Tbilisi. Porter, D. (1991). Affective factors in the assessment of oral interaction: Gender and status. In S. Anivan (Ed.), Current developments in languge testing. Anthology Series 25. Paper presented at the Regional Language Centre Seminar on Language Testing and Language Programme Evaluation (April 9–12, 1990). Porter, D., & Shen, S. H. (1991). Sex, status and style in the interview. The Dolphin, 21, 117–128. Porter, P. (1986). How learners talk to each other: Input and interaction in task-centred discussions. In R. R. Day (Ed.), Talking to learn: Conversation and second language acquisition (pp. 35–56). Rowley, MA: Newbury House. Poulisse, N. (1990). The use of compensatory strategies by Dutch learners of English: Dordrecht: Foris.

254

References

Poulisse, N. (1999). Slips of the tongue: Speech errors in first and second language production. Amsterdam: John Benjamins. Poulisse, N., & Bongaerts, T. (1994). First language use in second language production. Applied Linguistics, 15(1), 36–57. Poulisse, N., Bongaerts, T., & Kellerman, E. (1990). The use of compensatory strategies by Dutch learners of English. Enschede: Sneldruk. Puig-Mayenco, E., & Marsden, H. (2018). Polarity-item anything in L3 English: Where does transfer come from when the L1 is Catalan and the L2 is Spanish? Second Language Research, 34, 487–515. Qian, X. (2009). Codeswitching in teacher talk of primary English teachers in China. In B. Beaven (Ed.), IATEFL 2008: Exeter conference selections (pp. 101–102). Canterbury: IATEFL. Quián, D. D. (1999). Assessing the roles of depth and breadth of vocabulary knowledge in reading comprehension. Canadian Modern Language Review, 56, 282–308. Ramírez-Verdugo, D. (2006). A study of intonation awareness and learning in non-native speakers of English. Language Awareness, 15(3), 141–159. https://doi.org/10.2167/la404.0. Rast, R. (2010). The role of linguistic input in the first hours of adult language learning. Language Learning, 60(2), 64–84. https://doi.org/10.1111/j.1467-9922.2010.00601.x. Renou, J. (2001). An examination of the relationship between metalinguistic awareness and second-language proficiency of adult learners of French. Language Awareness, 10(4), 248–267. Reyes, M. (2011). ¡Felicidades! Felicitar como acto de habla. Decires. Revista del Centro de Enseñanza para Extranjeros, 13, 71–88. Ricciardelli, L. A. (1992). Bilingualism and cognitive development in relation to threshold theory. Journal of Psycholinguistic Research, 21, 301–316. Ringbom, H. (1986). Cross-linguistic influence and the foreign language learning process. In E. Kellerman & M. Sharwood Smith (Eds.), Crosslinguistic Influence in Second Language Acquisition (pp. 150–162). Oxford: Pergamon. Ringbom, H. (1987). The role of the first language in foreign language learning. Clevedon: Multilingual Matters. Ringbom, H. (2001). Lexical transfer in L3 production. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Cross -linguistic influence in third language acquisition: Psycholinguistic perspectives (pp. 59–68). Clevedon: Multilingual Matters. Ringbom, H. (2006). The importance of different types of similarity in transfer studies. In J. Arabski (Ed.), Cross-linguistic influences in the second language lexicon (pp. 36–45). Clevedon: Multilingual Matters. Ringbom, H. (2007). Cross-linguistic similarity in foreign language learning. Clevedon: Multilingual Matters. Ringbom, H., & Jarvis, S. (2009). The importance of cross-linguistic similarity in foreign language learning. In M. H. Long & C. J. Doughty (Eds.), The handbook of language teaching (pp. 107–118). Chichester: Wiley-Blackwell. Rocca, S. (2007). Child second language acquisition. Amsterdam: Benjamins. Rogerson, P. (2011). English phonology and pronunciation teaching. Great Britain: Continuum International Publishing Group. Roothooft, H. (2012). Developing a methodology for researching the effects of corrective feedback on the acquisition of the past tense: A pilot study. ELIA, 12, 211–242. Rose, K. R. (2000). An exploratory cross-sectional study of interlanguage pragmatic development. Studies in Second Language Acquisition, 22(1), 27–67. Rothman, J., & Cabrelli Amaro, J. (2010). What variables condition syntactic transfer? A look at the L3 initial state. Second Language Research, 26, 189–218. Rothman, J. (2010). On the typological economy of syntactic transfer: Word order and relative clause high/low attachment preference in L3 Brazilian Portuguese. International Review of Applied Linguistics, IRAL, 48(2/3), 245–273.

References

255

Rothman, J. (2011). L3 syntactic transfer selectivity and typological determinacy: The typological primacy model. Second Language Research, 27, 107–127. https://doi.org/10.1177/ 0267658310386439. Rothman, J. (2015). Linguistic and cognitive motivations for the typological primacy model (TPM) of third language (L3) transfer: Timing of acquisition and proficiency considered. Bilingualism: Language and Cognition, 18(2), 179–190. Rothman, J., & Cabrelli Amaro, J. (2010). What variables condition syntactic transfer? A look at the L3 initial state. Second Language Research, 26, 189–218. https://doi.org/10.1177/ 0267658309349410. Rutherford, W. (1988). Grammatical theory and L2 acquisition: A brief overview. In S. Flynn & W. O’Neil (Eds.), Linguistic theory in second language acquisition (pp. 404–417). The Netherlands: Kluwer Academic Publishers. Sabaté i Dalmau, M., & Curell i Gotor, H. (2007). From “Sorry very much” to “I'm ever so sorry”: Acquisitional patterns in L2 apologies by Catalan learners of English. Intercultural Pragmatics, 4, 287–315. Safronova, E., & Mora, J. C. (2012). Acoustic and phonological memory in L2 vowel perception. In S. Martín-Alegre, M. Moyer, E. Pladevall, & S. Tubau (Eds.), At a time of crisis: English and American studies in Spain (pp. 384–390). Barcelona: Departament de Filologia Anglesa i de Germanística, Universitat Autònoma de Barcelona. Saito, M. (1985). Some asymetries in Japanese and their theoretical implications. Doctoral dissertation. MIT. Salazar Campillo, P. (2006). Focus on form tasks and the provision of corrective feedback. Ibérica, 11, 123–138. Sánchez, L. (2003). Cross-linguistic influence and age. Paper presented at the Third International Conference on Third Language Acquisition and Trilingualism. Tralee, Eire-Ireland. Sánchez, L. (2011a). Crosslinguistic influence in third language acquisition: English after German in Spanish/Catalan learners in an instructed setting (Unpublished Ph.D. dissertation). University of Barcelona (European Doctorate Mention). Sánchez, L. (2011b). ‘Luisa and Perdito’s dog will the breakfast eat’: Interlanguage transfer and the role of the second language factor. In G. De Angelis & J.-M. Dewaele (Eds.), New trends in crosslinguistic influence and multilingualism research (pp. 86–104). Clevedon: Multilingual Matters. Sánchez, L. (2015). Can a background language alter the path of acquisition of verb placement in a multilingual context? Evidence from a longitudinal study. In G. De Angelis, U. Jessner, & M. Krésic (Eds.), Crosslinguistic influence and crosslinguistic interaction in language learning (pp. 125–141). Clevedon: Multilingual Matters. Sánchez, L., & Bardel, C. (2017). Transfer from an L2 in third language learning: A study on L2 proficiency. In T. Angeloviska & A. Hahn (Eds.), L3 syntactic transfer: Models, new developments and implications (pp. 223–250). Amsterdam/ Philadelphia: John Benjamins. Sánchez, L., & Jarvis, S. (2008). The use of picture stories in the investigation of crosslinguistic influence. TESOL Quarterly, 42(2), 329–333. Sánchez-Benedito, F. (1994). Manual de pronunciación inglesa comparada con la española. Madrid: Alhambra Longman. Sanz, C. (2000). Bilingual education enhances third language acquisition: Evidence from Catalonia. Applied Psycholinguistics, 21(1), 23–44. Schachter, J. (1974). An error in error analysis. Language Learning, 24, 205–214. Schmitt, N., Schmitt, D., & Clapham, C. (2001). Developing and exploring the behavior of two new versions of the vocabulary level test. Language Testing, 18(1), 55–88. Schumann, J. (1979). The acquisition of English negation by speakers of Spanish: A review of the literature. In R. Andersen (Ed.), The acquisition and use of Spanish and English as first and second languages (pp. 3–32). Washington DC: TESOL. Schwartz, B. D., & Sprouse, R. A. (1996). L2 cognitive states and the full transfer/full access model. Second Language Research, 12, 40–72. https://doi.org/10.1177/026765839601200103.

256

References

Searle, J. (1976). A classification of illocutionary acts. Language in Society, 5(1), 1–23. Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10, 209–231. Serra, C. (2007). Assessing CLIL at primary school: A longitudinal study. International Journal of Bilingual Education and Bilingualism, 10(5), 582–602. Sharwood Smith, M., & Kellerman, E. (1986). Crosslinguistic influence in second language acquisition: An introduction. In E. Kellerman & M. Sharwood Smith (Eds.), Crosslinguistic influence in second language acquisition (pp. 1–9). New York: Pergamon Press. Sharwood-Smith, M., & Kellerman, E. (1989). The interpretation of second language output. In H. W. Dechert & M. Raupach (Eds.), Transfer in language production (pp. 217–235). Norwoord, NJ: Ablex. Simsek, S. (2006). Third language acquisition: Turkish-German bilingual students’ acquisition of English word order in a German educational setting. Münster: Waxmann. Singleton, D. (1987). Mother and other tongue influence on learner French. Studies in Second Language Acquisition, 9, 327–346. https://doi.org/10.1017/S0272263100006719. Singleton, D. (2016). Cross-lexical interaction and the structure of the mental lexicon. In L. Yu & T. Odlin (Eds.), New perspectives on transfer in second language learning (pp. 51–62). Bristol: Multilingual Matters. Singleton, D., & Ó Laoire, M. (2006a). Psychotypologie et facteur L2 dans l’influence translexicale. Une analyse de l’influence de l’anglais et de l’irlandais sur le français L3 de l’apprenant. AILE, 24, 101–117. Singleton, D., & Ó Laoire, M. (2006b). Psychotypology and the ‘L2 factor’ in cross-lexical interaction: An analysis of English and Irish influence in learner French. In M. Bendtsen, M. Björklund, C. Fant & L. Forsman (Eds.), Språk, Lärande och Utbildning i sikte (pp. 191– 205). Vasa: Faculty of Education, Åbo Akademi. Skolverket. (2016). Språkval i grundskolan. Accessed June 30, 2017, at https://www.skolverket. se/skolutveckling/studie-och-yrkesvagledning/val-och-vagledning/syv-grundskola/sprakval-1. 209878. Slabakova, R. (2017). The scalpel model of third language acquisition. International Journal of Bilingualism, 21, 651–665. https://doi.org/10.1177/1367006916655413. Slabakova, R., & García Mayo, M. P. (2015) The L3 syntax–discourse interface. Bilingualism: Language and Cognition,18(2):208–226. Slabakova, R., & García Mayo, M. P. (2017) Testing the current models of third language acquisition. In T. Angeloviska & A. Hahn (Eds.), L3 syntactic transfer: Models, new developments and implications (pp. 64–84). Amsterdam/Philadelphia: John Benjamins. Soto-Faraco, S., Sebastián-Gallés, N., & Cutler, A. (2001). Segmental and suprasegmental mismatch in lexical access. Journal of Memory and Language, 45(3), 412–432. https://doi.org/ 10.1006/jmla.2000.2783. Spada, N. (2015). SLA research and pedagogy: Misapplications and questions of relevance. Language Teaching, 48, 69–81. Spada, N. (2018). The relationship between SLA research and language education. In Plenary talk given at the 1st ELE conference: Exploring Language Education. Global and Local Perspectives. Stockholm, June 18–20, 2018. Spencer-Oatey, H. (2002). Managing rapport in talk: Using rapport sensitive incidents to explore the motivational concerns underlying the management of relations. Journal of Pragmatics, 34, 529–545. Spencer-Oatey, H. (2005). Rapport management theory and culture. Intercultural Pragmatics, 2(3), 335–346. Storch, N., & Aldosari, A. (2010). Learners’ use of first language (Arabic) in pair work in an EFL class. Language Teaching Research, 14(4), 355–375. Storch, N., & Ali, A. (2010). Learners’ use of first language (Arabic) in pair work in an EFL class. Language Teaching Research, 14, 355–375.

References

257

Svensson, G. (2009). Diskurspartiklar hos ungdomar i mångspråkiga miljöer i Malmö [Discourse particles among youth in multilingual environments in Malmö] Doctoral dissertation. Lund: Lund University. Swain, M. (2000). The output hypothesis and beyond: Mediating acquisition through collaborative dialogue. In J. Lantolf (Ed.), Sociocultural theory and second language learning (pp. 97–114). Oxford: Oxford University Press. Swain, M., & Lapkin, S. (2000). Task-based second language learning: The uses of the first language. Language Teaching Research, 4(3), 251–274. Swain, M., & Lapkin, S. (2013). A Vygotskian sociocultural perspective on immersion education: The L1/L2 debate. Journal of Immersion and Content-Based Education, 1, 101–129. Swan, M. (1997). The influence of the mother tongue on second language vocabulary acquisition and use. In N. Schmitt & M. McCarthy (Eds.), Vocabulary, description, acquisition and pedagogy (pp. 156–180). Cambridge: Cambridge University Press. Swanson, B., & Charniak, E. (2014). Data driven language transfer hypotheses. In Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics (Vol. 2: Short Papers, pp. 169–173). Tabernero, C. (2008). Disponibilidad léxica y contacto de lenguas. Oinhenart: cuadernos de lengua y literature, 23, 545–565. Taft, M., Castles, A., Davis, C., Lazendic, G., & Nguyen-Hoan, M. (2008). Automatic activation of orthography in spoken word recognition: Pseudohomograph priming. Journal of Memory and Language, 58(2), 366–379. http://doi.org/http://dx.doi.org/10.1016/j.jml.2007.11.002. Taponen, S. (2014). “Hey, What’s Up, Dude?” Greetings in email messages in L2 English and L2 Swedish. University of Jyväskylä, Department of Languages. Tarone, E. (1977). Conscious communication strategies in interlanguage: A progress report. In H. D. Brown, C. A. Yorio, & R. C. Crymes (Eds.), On TESOL 77: Teaching and learning english as a second language: Trends in research and practice (pp. 194–203). Washington D. C.: TESOL. Tavakol, M., & Jabbari, A. A. (2014). The evolution of perspectives on the role of transfer during multilingual development. International Journal of Research Studies in Language Learning, 3, 105–120. Tedick, D. J., & Wesely, P. M. (2015). A review of research on content-based foreign/second language education in US K-12 contexts. Language, Culture and Curriculum, 28(1), 25–40. Tetreault, J., Blanchard, D., Cahill, A., & Chodorow, M. (2012). Native tongues, lost and found: Resources and empirical evaluations in native language identification. In Proceedings of the 24th International Conference on Computational Linguistics (COLING), 2585–2602. Thomas, J. (1988). The role played by metalinguistic awareness in second and third language learning. Journal of Multilingual and Multicultural Development, 9, 235–246. Thomas, J. (1992). Metalinguistic awareness in second-and third-language learning. In R. Harris (Ed.), Cognitive processing in bilinguals (pp. 531–545). Amsterdam: North Holland. Thomason, S. G., & Kaufman, T. (1988). Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Thomson, R. I. (2012). Improving L2 listeners’ perception of English vowels: A computer-mediated approach. Language Learning, 62(4), 1231–1258. Thomson, R. I., & Derwing, T. M. (2016). Is phonemic training using nonsense or real words more effective. In J. Levis, H. Le, I. Lucic, E. Simpson, & S. Vo (Eds.), Proceedings of the 7th Pronunciation in Second Language Learning and Teaching Conference (pp. 88–97). Ames, IA: Iowa State University. Tognini, R. (2008). Interaction in languages other than English classes in Western Australian primary and secondary schools: Theory, practice and percentions. Doctoral Thesis, Edith Cowan University. Retrieved from https://ro.ecu.edu.au/theses/26. Tognini, R., & Oliver, R. (2012a). L1 use in primary and secondary foreign language classrooms and its contribution to learning. In E. Alcón Soler & M. P. Safont Jordá (Eds.), Discourse and learning across L2 instructional contexts (pp. 53–78). Amsterdam, The Netherlands: Rodopi.

258

References

Tognini, R., & Oliver, R. (2012b). L1 use in primary and secondary foreign language classrooms and its contribution to learning. In E. Alcón Soler & P. Safont-Jordà (Eds.), Language learners’ discourse in instructional settings (pp. 53–78). Amsterdam: Rodopi. Tucker, R. G. (1998). A global perspective on multilingualism and multilingual education. In J. Cenoz & F. Genesee (Eds.), Beyond bilingualism: Multilingualism and multilingual education (pp. 3–15). Clevedon: Multilingual Matters. Turnbull, M., & Arnett, K. (2002). Teachers' uses of the target and first languages in second and foreign language classrooms. Annual Review of Applied Linguistics, 22, 204–218. UCLES. (2001). The Oxford quick placement test. Oxford University Press, University of Cambridge & Association of Language Testers in Europe. Ullman, M. T. (2015). The declarative/procedural model: A neurobiologically motivated theory of first and second language. In B. VanPatten & J. Williams (Eds.), Theories in second language acquisition: An introduction (2nd ed., pp. 135–158). New York: Routledge. Ullman, M. T. (2016). The declarative/procedural model: A neurobiological model of language learning, knowledge, and use. In G. Hickok & S. L. Small (Eds.), Neurobiology of language (pp. 953–968). San Diego, CA: Elsevier. Ullman, M. T., & Lovelett, J. T. (2018). Implications of the declarative/procedural model for improving second language learning: The role of memory enhancement techniques. Second Language Research, 34, 39–65. Vainikka, A., & Young-Scholten, M. (1996). Gradual development of L2 phrase structure. Second Language Research, 12, 7–39. Van Leussen, J. W., & Escudero, P. (2015). Learning to perceive and recognize a second language: The L2LP model revised. Frontiers in Psychology, 6, 1–12. https://doi.org/10.3389/fpsyg. 2015.01000. Vanderheijden, V. (2010). “Is that English I hear over there?” Rethinking the prohibition on L1 use in foreign language learning. TPFLE-Texas Papers in Foreign Language Education, 14(1), 3–17. Victori, M., & Tragant, E. (2003). Learner strategies: A cross-sectional and longitudinal study of primary and high-school EFL learners. In M. P. García Mayo & M. L. García Lecumberri (Eds.), Age and the acquisition of English as a foreign language (pp. 182–209). Clevedon: Multilingual Matters. Viladot, J., & Celaya, M. (2007). ``How do you say preparar?'': L1 use in EFL oral production and task-related differences. In M. Losada, P. Ron, S. Hernández, & S. Casanova (Eds.), Proceedings of the 30th international AEDEAN conference. Huelva: Universidad de Huelva. CD format. Vroomen, J., & De Gelder, B. (1995). Metrical segmentation and lexical inhibition in spoken word recognition. Journal of Experimental Psychology, 21(1), 98–108. Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press. Walker, R. (2010). Teaching the pronunciation of English as a lingua franca. Oxford: Oxford University Press. Wang, L. M. (2008). EFL 学习者习得交际策略的性别差异研究 [A study of gender differences in communication strategies by EFL learners]. Foreign Languages and Their Teaching, 8, 37–41. Wang, X., & Munro, M. J. (1999). The perception of English tense-lax vowel pairs by native Mandarin speakers: The effect of training on attention to temporal and spectral cues. In J. J. Ohala (Ed.), Proceedings of the 14th International Congress of Phonetic Sciences (Vol. 3, pp. 125–128). Berkeley, CA: University of California. Wannaruk, A. (2003). Communication strategies employed by EST students. Studies of Language and Language teaching, 12, 1–18. Wanrooij, K., Escudero, P., & Raijmakers, M. E. J. (2013). What do listeners learn from exposure to a vowel distribution? An analysis of listening strategies in distributional learning. Journal of Phonetics, 41(5), 307–319. https://doi.org/10.1016/j.wocn.2013.03.005.

References

259

Weber, A., & Cutler, A. (2004). Lexical competition in non-native spoken-word recognition. Journal of Memory and Language, 50(1), 1–25. Weinreich, U. (1953). Languages in contact. The Hague: Mouton. Wells, G. (1999). Using L1 to master L2: A response to Antón and DiCamilla's ‘Socio-cognitive functions of L1 collaborative interaction in the L2 classroom’. The Modern Language Journal, 83(2), 248–254. Westergaard, M., Mitrofanova, N., & Mykhaylyk, R. (2017). Crosslinguistic influence in the acquisition of a third language: The Linguistic Proximity Model. International Journal of Bilingualism, 21(6), 666–682.https://doi.org/10.1177/1367006916648859. White, J. (1998). Getting the learners’ attention: A typographical input enhancement study. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 85–113). Cambridge: Cambridge University Press. White, L. (1989). Universal grammar and second language acquisition. Amsterdam/Philadelphia: John Benjamins. Williams, S., & Hammarberg, B. (1998). Language switches in L3 production: Implications for a polyglot speaking model. Applied Linguistics, 19, 295–333. https://doi.org/10.1093/applin/19. 3.295 (Reprinted and revised In B. Hammarberg (Ed.). (2009), Processes in third language acquisition, 28–73. Edinburgh: Edinburgh University Press). Wode, H. (1978). Developmental sequences in naturalistic SLA. In E. Hatch (Ed.), Second language acquisition: A book of readings (pp. 101–117). Rowley, MA: Newbury House. Wood, D., Bruner, J., & Ross, G. (1976). The role of tutoring in problem solving. Journal of Child Psychology and Psychiatry, 17, 89–100. Wrembel, M. (2012). Foreign accentedness in third language acquisition. In J. Cabrelli- Amaro, S. Flynn, & J. Rothman (Eds.), Third language acquisition in adulthood (pp. 281–310). Amsterdam: John Benjamins. Yazawa, K., Whang, J., Kondo, M., & Escudero, P. (2019). Language-dependent cue weighting: An investigation of perception modes in L2 learning. Second Language Research, 0267658319832645. Yip, V., & Matthews, S. (2000). Syntactic transfer in a Cantonese-English bilingual child. Bilingualism: Language and Cognition, 3(3), 193–208. Yoshida, M. T. (2018). Choosing technology tools to meet pronunciation teaching and learning goals. CATESOL Journal, 30(1), 195–212. Yule, G. (1996). Pragmatics. Oxford: Oxford University Press. Yule, G., & Macdonald, D. (1990). Resolving referential conflicts in L2 interaction: The effect of proficiency and interactive role. Language Learning, 40, 539–556. Zuengler, J., & Miller, E. R. (2006). Cognitive and sociocultural perspectives: Two parallel SLA worlds? TESOL Quarterly, 40, 35–58.

Index

A Accent foreign, 15, 23, 28, 218 Accuracy rate, 68, 73, 77, 79, 82, 219, 222–224, 226 Acoustic analysis, 13 properties, 5–7 signal, 2 similarity, 6, 7, 14 Activation dual, 3, 61 L1 and L2, 3, 60, 61, 63 simultaneous, 52, 60, 61 Affection expressions, 157 Allophone, 27, 32 Apologizing, 153, 160 Aspiration, 24, 27 Assertives, 151, 199 Assimilation, 8, 9, 16, 22, 188 Auditory cue, 5, 10, 13 Avoidance, 51, 62, 105, 139, 171, 172, 174, 212 Awareness cognate, 62 cross-linguistic, 62 metalinguistic, 53, 80, 82, 98, 102, 106, 114, 118, 227 orthographic, 62 phonetic, 62 strategy, 62 B Bilingual, 24, 25, 28, 36, 45–47, 53–55, 57, 61, 63, 67, 68, 79, 114, 176, 194, 222, 223

Bilingualism benefits of/advantage of, 50 Biliteracy, 175, 188 Borrowing, 50, 52, 174, 178, 186, 198 C Calque, 52, 56, 61, 131, 172, 174, 226 Categorisation, 9, 14, 16, 129, 130 Circumlocution, 131–133, 171, 178 Classification system, 216 Classifier, 223, 226, 230 Classroom environment, 199, 200 management, 200, 209 observation, 197, 202 procedure, 19, 200, 211 setting, 19, 195, 196, 211 Clauses embedded, 72, 77, 78 matrix, 77 Code-switching, 173, 197, 198 Cognate, 28, 33, 35, 37, 38, 46, 50, 51, 56, 57, 59, 60, 62, 63 Cognitive abilities, 15 demands, 140, 171, 193, 195, 197, 200 effort, 192 function, 193, 196, 197, 203, 213 learning, 15, 89, 107, 160, 192–196, 198, 200, 209, 210, 212, 213 process, 192, 193, 198, 199, 209, 212 processing, 194–196, 210 Collaborative activities, 193 dialogue, 194, 196–199, 212

© Springer Nature Switzerland AG 2019 M. J Gutierrez-Mangado et al. (eds.), Cross-Linguistic Influence: From Empirical Evidence to Classroom Practice, Second Language Learning and Teaching, https://doi.org/10.1007/978-3-030-22066-2

261

262 teaching, 195, 209, 210, 213 Collocation, 63 Commissives, 151 Communication strategy, 51, 122, 174, 175, 182, 185–188 Competence cross linguistic influence (CLI), 60, 63 lexical, 62, 63 strategic, 62 Computational linguistics, 216, 217 Computer-automated, 223, 231 Congratulating, 150–154, 156–159, 161–163 Consonant cluster, 32, 33, 35–38, 43, 44, 46 Content Teacher (CT), 202, 203, 205–211 Contrastive Analysis Hypothesis (CAH), 104 Cooperative, 195, 199, 205, 210, 213 Corpus, 216, 219, 226, 231 Co-teaching, 194, 195, 201, 202, 210, 211 Cross-Linguistic Influence (CLI), 2, 3, 13, 14, 18, 20, 22, 47, 50–61, 63, 66–68, 72, 74, 79–81, 89–91, 93–99, 102, 103, 105, 106, 113, 122–126, 129–132, 134–140, 212, 216–219, 222, 231 Cue-weighting, 5, 10–13, 16 Cumulative Enhancement Model (CEM), 102, 107, 108, 114–117 D Declarative/Procedural (D/P) (Model), 109 Declaratives, 67, 109, 110, 118, 119, 151, 200 Diphthong, 25–27, 36 Directives, 151 Discourse completion test (DCT), 152, 155 markers, 129, 186, 196–198 unplanned, 173, 188, 198, 212 Discrimination, 9, 10, 14–16, 61 Discursive, 22, 104 Distributional learning, 10, 11, 13, 19, 20 E English as a Foreign Language (EFL), 15, 23, 24, 31, 40, 45, 46, 53–55, 58, 61, 73, 82, 122, 123, 126, 127, 132, 150, 153, 154, 159, 161–163, 171, 176, 195–197, 201, 223 Explicit L1 Use, 122–125, 129, 130, 132, 133, 139, 140 F Failed Functional Features Hypothesis (FFH), 105

Index First Language (L1), 2, 5–18, 20–25, 45–47, 50–56, 59–63, 66–68, 72–75, 79–82, 88–92, 94–97, 102–120, 122–124, 126, 129–140, 150, 153–156, 159–163, 172, 173, 176, 177, 181, 184, 187, 192–203, 206–213, 216–231 Fluency, 23, 160 Foreignizings, 171, 172, 174, 178, 186 Foreign Language (FL), 19, 21, 23, 28, 45, 51, 53, 54, 59, 61–63, 81, 82, 88–92, 96–98, 102, 103, 118–120, 124, 150, 153–155, 160, 162, 163, 170, 173, 175, 176, 192–198, 200–202, 208–210, 212, 213 Formal language learning, 102, 103 Fossilization, 140 Full Transfer /Full Access (FT/FA) Hypothesis, 105, 110 Functional phonology model, 4 G Gender, 124, 131, 135, 136, 155, 171, 187 Gradual Learning Algorithm (GLA), 4, 10, 11, 13 Grammaticality filter, 68, 80 Grammaticality Judgement Task (GJT), 68, 73, 74, 76, 79–82 Grapheme, 25 Gratitude, 163 Greetings, 152 H H-deletion, 32, 33, 40 Higher-level texts/ essays, 224, 225 High Variability Phonetic Training (HVPT), 18, 19 Human judges, 217–223, 226–231 I Illocutionary Force Indicating Device (IFID), 152, 153, 157 Implicit competence, 116, 118 Inhibition, 54, 62, 66 Inhibitory control /strategy, 54, 61, 63 Initial state/stage, 4, 5, 7, 8, 10, 14, 18, 79, 81, 105, 109–111, 116 Intelligibility, 23, 25 Intensifiers, 162 Interaction, 15, 38, 39, 50, 52, 102, 104, 121, 122, 124–126, 128, 129, 133, 139, 140, 155, 162, 163, 183, 193–200, 202, 203, 205–213 Interlanguage pragmatics (ILP), 150

Index International Corpus of Learner English (ICLE), 223, 226 Intervention, 211

263 Motivation, 15, 115, 116, 123, 125, 133, 139, 200 Multilinguals, 53, 66, 80, 81, 87–91, 94, 96, 98, 99, 102–105, 108, 111, 113, 115, 117, 120, 173, 194, 223

L Level-matched, 140 Lexical access/accessibility, 52, 55, 60, 63 availability task, 50, 52, 54–56, 59–61, 63 creation, 55, 57–60, 63 level, 50, 53, 59, 60, 98, 106 transfer, 22, 50–54, 63, 105, 116, 218 Lexical cross-linguistic influence, 2, 50, 53–57, 59–61, 63, 91, 93–98, 129, 130, 136, 140 L1 Identification, 217–224, 226–231 L1 Indicators, 219, 220, 223, 224, 226, 228, 229 Linear discriminant analysis, 216, 223, 226 Linguistic Proximity Model (LPM), 102, 113, 116, 118 Literal translation, 52, 56–60, 63, 122, 171, 178 L2 Status Factor hypothesis (L2SFH), 102, 108

O Optimality Theory (OT), 4, 7, 8, 10, 13 Orthography/Orthographic interference, 24, 25, 28, 32–35, 37–40, 43–46 transfer, 24, 25, 29, 31–33, 35, 37, 38, 40, 42, 45, 46 transparent/transparency, 25, 31, 38, 45, 46 Overgeneralisation, 123, 131, 135, 136, 140 Over-identification, 221, 226

M Machine-learning, 216–219, 222, 223, 226–231 Matched-pairs, 132, 140 Memory declarative, 109, 118, 119 procedural, 109, 118 Meta-classifiers, 230 Metacomments, 172, 173, 197 Metalinguistic awareness, 53, 80, 82, 90, 98, 102, 106, 114, 118, 120, 227 knowledge, 80, 82, 90, 98, 102, 104, 109, 110, 114, 116, 118–120 Metapragmatic, 162 Metatalk, 193, 197, 199, 202, 203, 205, 207–209 Misspellings, 52, 222 Monolingual, 6, 53, 54, 57, 58, 61, 63, 67, 68, 72, 73, 79–82, 113, 128, 133, 136, 138, 140, 192, 194 Monophthong orthographic, 25, 35, 38 Morphology/Morphological, 22, 60, 111, 130, 174, 218 Morphosyntactic/Morphosyntax, 55, 110, 111, 130, 196 Mother–Tongue, 49, 54, 88, 89, 120, 125, 127

P Paraphrase/Paraphrasing, 171, 172, 174, 178 Parts of Speech (POS), 216 Pedagogical intervention, 82, 140 Perceptual Assimilation Model (PAM), 2 Performance CLI (PCLI), 60 Phoneme, 2, 9, 24, 27, 32, 125 Phoneme-to-grapheme correspondence, 25 Phonetic spelling, 53, 55, 57–61, 63 Phonological Interference, 25, 32, 33, 35–38, 40, 43, 44, 46 transfer, 22, 24, 31, 32, 35, 40, 42 Picture stories, 125 Pitch, 15 Politeness, 151–153 Possessives, 131, 135 Postposition, 69 Pragmatic(s) awareness, 162 Preposition, 31, 39, 69, 74, 130–132, 135, 139 Previously known languages, 172–175, 186–188 Processing strategy, 11, 19 Pronunciation, 17–19, 22–25, 29, 31–33, 35–38, 40, 44–47, 61, 62, 110, 119 Psychotypology, 68, 106, 111, 114

N Narration, 18, 172, 177, 179, 185, 187, 196, 198 Native Language Identification (NLI), 216–219, 221–223, 230, 231 Negotiation, 121, 122, 138 N-grams, 216 Non-native, 3, 6–9, 16, 47, 103–107, 109, 110, 113, 114, 116, 119, 130, 151

264 R Raters, 217, 218 Receptive vocabulary, 171, 172, 175, 177, 179, 180, 184, 186, 187 Repetitions, 119, 122, 123, 139, 155 S Salience, 219 Scaffolding, 51, 194, 197, 209, 212, 213 Scalpel model, 102, 111, 112, 115, 116, 118 Schwa, 27, 31, 33, 46 Second language acquisition, 2, 66, 87, 102, 121, 170 perception Model, 3, 4 Self-regulation, 197, 202, 203, 205, 208, 209 Semantic extensions, 52, 63 features, 51, 52, 57 Sociolinguistic, 22, 116, 150, 175 Sound, 2, 3, 5, 6, 8–10, 15, 17–19, 22, 24–28, 30–33, 35, 36, 38, 40, 45–47, 60, 61 Speech act, 150–154, 156, 157, 159–163 Speech Learning Model (SLM), 2, 8, 9 Speech/sound perception, 2–4, 6, 11, 14 Spelling, 25–27, 31, 33, 35, 36, 38, 45, 53, 61, 62 Stochastic Optimality Theory, 3, 4 Storytelling activity, 127, 179 Strategy analytic, 170, 171, 177, 178, 181–185, 188 appeals for assistance, 172–174, 177, 181–184, 186, 187, 197 avoidance, 172, 174 communication, 51, 122, 170–172, 174, 175, 180, 182, 183, 185–188, 197 communicative, 60 compensatory, 54, 60, 62, 197–199, 212 holistic, 170, 171, 177, 178, 182–188 interactional, 170, 172, 173, 181, 183, 185, 192, 197–200, 202–204, 207, 208, 212 knowledge-based, 172 language based, 177 learning, 20, 53, 62, 198 metacognitive, 205, 210, 211 organisational, 205, 209 social/affective, 203, 206, 210, 211 subject-verb inversion, 135 Stress, 2, 23, 24, 28, 32, 33, 35–38, 42, 45, 46, 90, 96 Stressed/unstressed syllables, 28, 33 Structural errors, 218 transfer, 122–124, 132–135, 139, 140, 219

Index Subordinate, 67, 108, 170, 178 Superlative, 31, 38, 39 Superordinate, 170, 178 Suprasegmental aspects, 24 features, 45 Syllable, 3, 28, 32, 33, 35, 37 Syntactic transfer, 67, 68, 105–108 T Task communicative, 52, 172, 194, 197 complexity, 123, 139 repetition, 123, 139, 155 spot-the-difference, 127 Taxonomy, 152, 170 Teachability, 172, 188 Third language learning, 98 Tip-Of-the-Tongue (TOT), 53 Transfer facilitative, 72, 79, 81 lapse, 181, 185 non-facilitative/negative, 66, 79, 105, 108, 111, 114–116, 118, 120 Translation direct /word-for-word, 130 direct/word-for-word, 222, 226, 228 task, 73, 75, 76, 78–80 Trilingualism, 176 Triphthong, 26, 27 Typological Primary Model (TPM), 102, 106, 107, 110–116, 118 Typology, 67, 90, 96, 107, 110, 111, 113, 115, 118 U Under-identification, 221 Universal Grammar (UG), 105 V Variability, 16, 18, 135 Verb second (V2), 67, 68 Vocabulary, 31, 39, 42, 50, 52, 54, 61–63, 66, 76, 81, 90, 97, 98, 109, 129, 177, 186, 198 Vocabulary Level Test (VLT), 177, 179, 184, 186 Voiced, 3, 27 Voiceless, 3, 27 Vowel epenthetic, 32, 33, 36–38, 40, 42–45

Index W Word order basic, 69, 72, 82 free, 67 Word recognition, 2, 3, 9, 11–14, 17–20 Written discourse completion task, 150, 152, 153, 155, 163 text / essay, 218, 220–222

265 Y Young learners, 113, 122, 125, 162, 175, 187 Z Zone of Proximal Development (ZPD), 193, 196