The effects of different instructional and cognitive variables on the acquisition of

grammatical gender by second language learners of French

Andrew Hyunmin Lee

Department of Integrated Studies in Education

(Concentration: Second Language Education)

McGill University, Montreal

June, 2018

A thesis submitted to McGill University in partial fulfillment of the requirements of

the degree of Doctor of Philosophy

Copyright © Andrew H. Lee, 2018

 ii

This doctoral dissertation is dedicated to

my parents and mentors

who made me who I am today.

 iii

Acknowledgements

Looking back at the past six years and writing these acknowledgments, I realized

how privileged I was as a graduate student at McGill University. I would like to take this

opportunity to acknowledge my mentors, colleagues, friends, students, and family

members who contributed various and important roles to my graduate studies including

this dissertation.

Without my doctoral supervisor, Dr. Roy Lyster, I would not have even thought

of pursuing a PhD. Roy is the one who discovered my talent as a researcher, which I had

not realized by myself, and nurtured the talent for me. He has always been my role

model, showing me remarkable examples as a scholar, a teacher, and a mentor. Whenever

I had good news, he was happier than me. He encouraged me and showed me directions

whenever I stumbled during my MA and PhD journeys. I thank you, Roy, for your

endless and devoted guidance and support. I will not forget them.

Dr. Murray J. Munro at Simon Fraser University and Dr. Pavel Trofimovich at

Concordia University guided me as my doctoral committee members. Based on my

candidacy papers, they provided me with brilliant ideas and encouraged me to explore

more second language phonetics. In addition, they helped me develop and strengthen my

dissertation project. I would also like to acknowledge Dr. Kazuya Saito at Birkbeck,

University of London. He introduced me to classroom-based pronunciation instruction

and encouraged me to pursue my research interests with full enthusiasm. I thank Dr.

Marnie Reed at Boston University for agreeing to be my external reviewer.

I would like to extend my appreciation to Dr. Anila Asghar, who shared with me

numerous scholarly dialogues about social, political, and religious issues around the
 iv

world. She challenged me (positively) and always inspired me to become a better thinker,

believer, and human being ultimately. I thank you, Ammiji, for taking me as your Korean

son. The time and delicious cuisine that I shared with you and Sami are unforgettable.

My dissertation project would not be possible without the teachers and students in

the McGill French Language Centre. Natallia Liakina, who is the director of the Centre,

was highly interested in my research project and let me conduct a classroom-based study

in the Centre. Alida Soucé, Chantal Creck, Kevin Papin, Marie-Josée Fortin, and Viviane

Kwan-Lock were generous enough to open their classrooms to me and allow me to

implement experimental components during their invaluable class hours. The McGill

Arts Multimedia Language Facility and student assistants helped me administer tests as

smoothly as possible without any technical issues. More importantly, it was my great

pleasure to work with the participating students who were curious about my study and

showed me their genuine interest.

I was very fortunate to work with the following research assistants who supported

me whenever I asked for help and whatever the help was. Hyunseon Shin, Jonathan

Provost, and Stephen Davis helped me prepare classroom and testing materials. Alexis

Stylianou, Bosco Bampembe, and Amanda Walker not only served as liaison between me

and the classrooms but also assisted me in conducting tests with the students. Data

analysis was supported by Adam Saucier, Emily Denney, Emmy Côté, Jean-Philippe

Caisse, Justine Clément, Noémie François-Haugrin, Ophélie Gauthier-Barrette, and

Pierre-Yves Bonin. I also appreciate Stéphane Rivard in LearningBranch for allowing me

to use his platform for this project.

 v

I am thankful to my fellow graduate students and friends who spent a McGill life

with me: Benjamin Gormley, Megumi Fujio, Misung Kwon, Reggie Gooch, Takashi

Oba, Wenrui Duan, and Yiwei Zhou. I am also grateful to Donghyun Kim and Hye-

Young Bang in the Department of Linguistics, together with whom we literally shared

tears of joy and pain as doctoral students. Without our detox hours every weekend, none

of us would have been able to finish our degrees!

During my graduate studies, I received the full support of the Korean community

in Montreal, including several community-based scholarships. Jade Seo and Yuli always

welcomed me to their house and offered me memorable times during the Korean holidays

(with the bonus of Jade’s marvelous Korean dishes). Working as a course lecturer at

McGill stimulated me intellectually and helped me to not forget where I came from. My

students truly deserve a mention.

Last but not least, I cannot express my gratitude enough for my family. My

parents never questioned what I had pursued, but showed me their constant love, support,

and trust, all of which made me who I am today. I thank my sister, Hyuna, for being with

my parents and looking after important family affairs and events alone during my long

absence.

This dissertation study was supported by a Doctoral Fellowship from Le Fonds de

Recherche du Québec–Société et Culture, a McGill Graduate Dean’s Award, an Insight

Development Grant (#430-2017-00372) awarded by the Social Sciences and Humanities

Research Council of Canada to Roy Lyster, and a Language Learning Dissertation Grant.
 vi

Abstract

Previous studies (Harley, 1998; Lyster, 2004; Lyster & Izquierdo, 2009; Warden,

1997) found that learners of French as a second language (L2) benefit from form-focused

instruction (FFI) targeting morphological awareness of noun endings (i.e., sublexical

cues) in the acquisition of French grammatical gender. A noteworthy finding from

previous studies, however, is that L2 learners developed an interlanguage strategy of

pronouncing French articles in an ambiguous manner as hybrid forms between un and

une and between le and la. Thus, the present study hypothesizes that, in order for L2

learners to demonstrate targetlike performance regarding French grammatical gender, FFI

targeting the pronunciation of gender-specific definite and indefinite articles as well as

sublexical cues should be implemented in classroom instruction. Moreover, given the

importance of L2 learners’ executive function (EF) skills in L2 acquisition (Darcy, Mora,

& Daidone, 2016; Kapa & Colombo, 2014; Linck, Osthus, Koeth, & Bunting, 2014), the

current study also predicts that the extent to which L2 learners benefit from FFI will be

mediated by their EF skills operationalized as inhibitory control, nonverbal visuospatial

working memory, and cognitive flexibility.

To examine the hypotheses, a quasi-experimental study was conducted in six

intact French L2 classrooms for university-level learners (N = 140) comprising three

instructional conditions (two classrooms per condition): (a) FFI on only sublexical cues

(n = 41); (b) FFI on both sublexical cues and pronunciation (n = 49); (c) control (n = 50).

Those in the two FFI conditions received six 80-minute instructional sessions targeting

grammatical gender and those in the control condition continued with their regular

French L2 program.
 vii

To measure the effects of the instructional treatments on the acquisition of French

grammatical gender, a pretest, an immediate posttest, and a delayed posttest were

administrated, each of which included grammatical judgment, text-completion, forced-

choice identification, read-aloud, picture-description, and article-noun

congruent/incongruent tasks. To measure their EF skills, the Simon Test, the Corsi

Block-Tapping Test, and the Wisconsin Card Sorting Test were administered at each of

the three testing times.

Results show that participants in both FFI conditions made significant gains on

the posttests in the grammatical judgment, text-completion, and article-noun

congruent/incongruent tasks. In the read-aloud and picture-description tasks, participants

receiving FFI on both sublexical cues and pronunciation attained significantly higher

scores compared to their pretest scores and to those of the control group on the posttests,

whereas participants receiving FFI only on sublexical cues did not demonstrate any

significant improvement over time on either of these tasks. The multiple regression

analyses confirmed that L2 phonological knowledge is a significant predictor of

improving participants’ accuracy regarding French grammatical gender.

In the forced-choice identification task, all participants, regardless of condition,

achieved maximum scores at all three testing times, meaning that participants did not

have any difficulty perceptually categorizing the sounds of un, une, le, and la.

Participants in the control condition made no significant improvement on any measures.

For participants in both FFI conditions, nonverbal visuospatial working memory

was a significant predictor of the learning gains in the grammatical judgment task, while

inhibitory control was a significant predictor of the gains made in the read-aloud task.
 viii

The current study sheds light on the importance of L2 phonological knowledge

and L2 pronunciation instruction in the acquisition of French grammatical gender in

addition to the roles of EF skills in L2 acquisition.

 ix

Résumé

Des études antérieures (Harley, 1998; Lyster, 2004; Lyster et Izquierdo, 2009;

Warden, 1997) ont révélé que les apprenants en classe de français langue seconde (L2)

bénéficiaient d’un enseignement centré sur la forme (ECF) ciblant une conscientisation

morphologique de la terminaison des noms, c’est-à-dire les indices sublexicaux, dans

l’acquisition des genres grammaticaux en français. Les travaux précédents, cependant,

ont fait ressortir que les apprenants en classe de français L2 développaient une stratégie

de prononciation ambigüe des articles, utilisant une forme hybride entre un et une et entre

le et la. Afin que les apprenants en français L2 démontrent la compétence souhaitée pour

le genre grammatical, cette étude formule ainsi l’hypothèse que l’ECF, visant la

prononciation spécifique de l’article défini et indéfini de même que les indices

sublexicaux, devrait être intégré à l’enseignement en classe. Plus encore, étant donné

l’importance des capacités de fonction exécutive chez les apprenants dans l’acquisition

de la L2 (Darcy, Mora et Daidone, 2016; Kapa et Colombo, 2014; Linck, Osthus, Koeth

et Bunting, 2014), cette étude prédit également que les gains issus de l’ECF seront

déterminés par les capacités de fonction exécutive, opérationnalisées par le contrôle

inhibitoire, la mémoire de travail non verbale et visuo-spatiale et la flexibilité cognitive.

Dans le but d’examiner ces hypothèses, une étude quasi-expérimentale a été

menée auprès de six classes intactes en français L2 pour des apprenants de niveau

universitaire (N = 140) comprenant trois conditions (deux classes par condition) : (a) ECF

ciblant les indices sublexicaux seulement (n = 41) ; (b) ECF ciblant les indices

sublexicaux et la prononciation (n = 49) ; (c) groupe témoin (n = 50). Les étudiants ayant

été soumis aux deux conditions de l’ECF ont reçu des séances d’enseignement de 80
 x

minutes ciblant le genre grammatical, tandis que le groupe témoin a suivi le programme

régulier de français L2.

Pour mesurer les effets des conditions de l’ECF sur l’acquisition du genre

grammatical en français, un pré-test, un post-test immédiat et un post-test différé ont été

administrés, chacun d’eux incluant des tâches de jugement grammatical, de complétion

de texte, de choix binaire, de lecture à voix haute, de description d’illustrations et

d’accord ou de désaccord entre article et nom. Dans l’objectif d’évaluer les capacités de

fonction exécutive, le test Simon, le test de Corsi et le test de classement de cartes du

Wisconsin ont été conduits pour les trois tests (pré-test, post-test immédiat et post-test

différé).

Les résultats montrent que les participants dans les deux groupes ayant reçu l’ECF

se sont améliorés à leurs post-tests, notamment pour les tâches de jugement grammatical,

de complétion de texte et d’accord ou de désaccord entre article et nom. En ce qui a trait

aux tâches de lecture à voix haute et de description d’illustrations, les participants ayant

bénéficié de l’ECF ciblant et les indices sublexicaux et la prononciation ont obtenu des

résultats significativement plus élevés à leur post-test que le groupe témoin, alors que les

participants ayant reçu l’ECF portant uniquement sur les indices sublexicaux n’ont pas

démontré de progrès significatif avec le temps dans les tâches de lecture à voix haute et

de description d’illustrations. Les analyses de régression multiples confirment que la

connaissance phonologique en L2 est un prédicteur important de l’amélioration du

participant quant à la précision du genre grammatical en français.

Concernant la tâche de choix binaire, tous les participants, peu importe leur

condition, ont atteint des résultats optimaux autant au pré-test qu’aux post-tests, ce qui
 xi

signifie que ceux-ci n’avaient pas de difficulté à catégoriser de manière perceptuelle les

sons un, une, le et la. Le groupe témoin n’a pas fait de progrès dans aucune des tâches.

Pour les participants soumis aux conditions de l’ECF, la mémoire de travail non

verbale et visuo-spatiale s’est avérée un prédicteur important des gains relatifs aux

apprentissages dans la tâche de jugement grammatical, tandis que le contrôle inhibitoire a

été révélateur des progrès réalisés dans la tâche de lecture à voix haute.

L’étude présente met en lumière l’importance de la connaissance phonologique et

de l’enseignement de la prononciation de la L2 dans l’acquisition du genre grammatical

en français, de même que les rôles exercés par les capacités de fonction exécutive dans

l’apprentissage de la L2.
 xii

Table of Contents

Acknowledgments.............................................................................................................. iii
Abstract .............................................................................................................................. vi
Résumé............................................................................................................................... ix
Table of Contents .............................................................................................................. xii
List of Tables ................................................................................................................... xiv
List of Figures .................................................................................................................. xvi

Chapter 1. Introduction ................................................................................................... 1

1. 1. Structure of the Dissertation .................................................................................... 3

Chapter 2. Background .................................................................................................... 5

2. 1. Motivation of the Present Study .............................................................................. 5
2. 2. French Grammatical Gender ................................................................................... 7
2. 3. Noun Endings as Predictors of French Grammatical Gender................................ 11
2. 4. Form-Focused Instruction on Second Language Learning.................................... 15
2. 4. 1. Definition and Components of Form-Focused Instruction ............................ 15
2. 4. 2. Effectiveness of Form-Focused Instruction on French Grammatical Gender 16
2. 5. Executive Functions Skills in Second Language Learning ................................... 19
2. 5. 1. Inhibitory Control........................................................................................... 19
2. 5. 2. Working Memory ........................................................................................... 21
2. 5. 3. Cognitive Flexibility ...................................................................................... 23
2. 6. Research Questions................................................................................................ 24

Chapter 3. Methodology ................................................................................................. 26

3. 1. Participants ............................................................................................................ 26
3. 2. Procedures ............................................................................................................. 29
3. 3. Target Noun Endings ............................................................................................. 32
3. 4. Instructional Sessions ............................................................................................ 33
3. 4. 1. Overview ........................................................................................................ 33
3. 4. 2. Condition 1: Form-Focused Instruction on Only Sublexical Cues (Classes 1
and 2) .............................................................................................................. 35
3. 4. 3. Condition 2: Form-Focused Instruction on Both Sublexical Cues and
Pronunciation (Classes 3 and 4) ..................................................................... 37
3. 4. 4. Condition 3: Control Condition (Classes 5 and 6) ......................................... 40
3. 5. Measures ................................................................................................................ 41
3. 5. 1. Overview ........................................................................................................ 41
3. 5. 2. Target Nouns .................................................................................................. 43
3. 5. 3. Grammatical Judgment Task .......................................................................... 44
3. 5. 4. Text-Completion Task.................................................................................... 44
3. 5. 5. Forced-Choice Identification Task ................................................................. 45
3. 5. 6. Read-Aloud Task............................................................................................ 45
3. 5. 7. Picture-Description Task ................................................................................ 45
3. 5. 8. Article-Noun Congruent/Incongruent Task ................................................... 47
3. 5. 9. Measures of Executive Function Skills .......................................................... 48
 xiii

3. 6. Data Preparation .................................................................................................... 49

3. 7. Summary................................................................................................................ 51

Chapter 4. Results ........................................................................................................... 53

4. 1. Pre-Analysis........................................................................................................... 53
4. 2. Effects of Two Form-Focused Instructional Conditions ....................................... 54
4. 2. 1. Statistical Model............................................................................................. 54
4. 2. 2. Grammatical Judgment Task .......................................................................... 56
4. 2. 3. Text-Completion Task.................................................................................... 58
4. 2. 4. Forced-Choice Identification Task ................................................................. 61
4. 2. 5. Read-Aloud Task............................................................................................ 62
4. 2. 6. Picture-Description Task ................................................................................ 66
4. 2. 7. Article-Noun Congruent/Incongruent Task ................................................... 68
4. 3. Variables Affecting the Performance of the Picture-Description Task ................. 70
4. 3. 1. Statistical Model............................................................................................. 70
4. 3. 2. Results ............................................................................................................ 71
4. 4. Roles of Executive Function Skills in Learning Gains.......................................... 73
4. 4. 1. Statistical Model............................................................................................. 73
4. 4. 2. Results ............................................................................................................ 74
4. 5. Summary................................................................................................................ 79

Chapter 5. Discussion ..................................................................................................... 82

5. 1. Differential Effects of Form-Focused Instruction on Only Sublexical Cues and on
Both Sublexical Cues and Pronunciation .............................................................. 82
5. 1. 1. Grammatical Judgment and Text-Completion Tasks ..................................... 83
5. 1. 2. Forced-Choice Identification and Read-Aloud Tasks .................................... 84
5. 1. 3. Picture-Description and Article-Noun Congruent/Incongruent Tasks ........... 86
5. 1. 4. Summary ........................................................................................................ 89
5. 2. Roles of Executive Function Skills in Second Language Development ............... 90
5. 3. Summary................................................................................................................ 92

Chapter 6. Conclusion .................................................................................................... 93

6. 1. Summary of the Current Study .............................................................................. 93
6. 2. Pedagogical Implications ....................................................................................... 95
6. 3. Future Directions ................................................................................................... 96

References ......................................................................................................................... 99

Appendices ...................................................................................................................... 112

Appendix A. Informed Consent Form for Participating Students ............................... 112
Appendix B. Instructional Materials for Form-Focused Instruction on Only Sublexical
Cues (e.g., Les Lettres Chinoises : 9) ..................................................... 114
Appendix C. Instructional Materials for Articulation-Based Instruction (Explicit
Phonetic Information) ............................................................................. 121
Appendix D. Instructional Materials for Articulation-Based Instruction (Notice-
Articulatory-Gestures and Mirror Activities) ......................................... 123
 xiv

List of Tables

Table 1. Form-Focused Instruction on French Grammatical Gender ............................... 17

Table 2. Background of Participating Students ................................................................ 27
Table 3. Summary of Target Noun Endings ..................................................................... 33
Table 4. Mean Percentage Scores and Standard Deviations (in parentheses) for the
Grammatical Judgment Task .............................................................................. 57
Table 5. Summary of Fixed Effects for the Statistical Model for the Grammatical
Judgment Task .................................................................................................... 58
Table 6. Mean Scores and Standard Deviations (in parentheses) for the Text-
Completion Task ................................................................................................. 59
Table 7. Summary of Fixed Effects for the Statistical Model for the Text-Completion
Task ..................................................................................................................... 60
Table 8. Mean Scores and Standard Deviations (in parentheses) for the Forced-Choice
Identification Task ............................................................................................. 61
Table 9. Summary of Fixed Effects for the Statistical Model for the Forced-Choice
Identification Task .............................................................................................. 62
Table 10. Mean Scores and Standard Deviations (in parentheses) for the Read-Aloud
Task ................................................................................................................... 63
Table 11. Summary of Fixed Effects for the Statistical Model for un and une ................ 64
Table 12. Summary of Fixed Effects for the Statistical Model for le and la .................... 65
Table 13. Mean Scores and Standard Deviations (in parentheses) for the Picture-
Description Task ............................................................................................... 66
Table 14. Summary of Fixed Effects for the Statistical Model for the Picture-
Description Task ............................................................................................... 67
Table 15. Mean Scores and Standard Deviations (in parentheses) for the Article-Noun
Congruent/Incongruent Task ............................................................................ 68
Table 16. Summary of Fixed Effects for the Statistical Model for the Article-Noun
Congruent/Incongruent Task ............................................................................ 69
Table 17. Linear Model of Predictors of the Performance of the Picture-Description
Task on the Pretest with 95% Confidence Intervals (in parentheses) ............... 71
Table 18. Linear Model of Predictors of the Performance of the Picture-Description
Task on the Immediate Posttest with 95% Confidence Intervals (in
parentheses)....................................................................................................... 72
Table 19. Linear Model of Predictors of the Performance of the Picture-Description
Task on the Delayed Posttest with 95% Confidence Intervals (in parentheses)
........................................................................................................................... 73
Table 20. Linear Model of Predictors of the Learning Gains of the Grammatical
Judgment Task with 95% Confidence Intervals (in parentheses) ..................... 75
Table 21. Linear Model of Predictors of the Learning Gains of the Text-Completion
Task with 95% Confidence Intervals (in parentheses) ..................................... 76
Table 22. Linear Model of Predictors of the Learning Gains of the Read-Aloud Task
with 95% Confidence Intervals (in parentheses) .............................................. 77
Table 23. Linear Model of Predictors of the Learning Gains of the Picture-Description
Task with 95% Confidence Intervals (in parentheses) ..................................... 78
 xv

Table 24. Linear Model of Predictors of the Learning Gains of the Article-Noun
Congruent/Incongruent Task with 95% Confidence Intervals (in parentheses)
........................................................................................................................... 79
 xvi

List of Figures

Figure 1. Components of proactive form-focused instruction (Ranta & Lyster, 2018,

p. 43) ................................................................................................................. 15
Figure 2. Example of a picture-description task ............................................................... 46
Figure 3. Boxplots of the scores of grammatical measures .............................................. 60
Figure 4. Boxplots of the scores of phonological measures ............................................. 65
Figure 5. Boxplots of the scores of grammatical + phonological measures ..................... 70
 1

Chapter 1

Introduction

In French, all nouns have grammatical gender. In the case of inanimate nouns,

they can be either masculine or feminine regardless of any semantic basis for gender

attribution (Sokolik & Smith, 1992). For instance, un chapeau is masculine, whereas une

casquette is feminine although both nouns indicate similar objects (i.e., a hat or a cap).

Previous studies (Clark, 1985; Karmiloff-Smith, 1979; van Heugten & Shi, 2009) found

that native speakers of French develop a powerful and implicit grammatical gender

system by the age of 3. However, learners of French as a second language (L2) have

difficulty mastering French grammatical gender in spite of intensive learning experiences

such as immersion contexts and of the high frequency of gender markers in linguistic

input (Harley, 1998; Lyster, 2004). In this regard, Harley (1998) argued that grammatical

gender is not something that L2 learners can learn incidentally. Carroll (1989) also

contended that L2 learners need mnemonic strategies and rules, which “could provide the

advanced learner not only with a reasonably accurate system but also with a mechanism

for guessing the gender of new items” (p. 580).

Given that noun endings are reliable predictors of gender attribution in French

(Tucker, Lambert, & Rigault, 1977; Tucker, Lambert, Rigault, & Segalowitz, 1968),

previous studies (Harley, 1998, Lyster, 2004; Lyster & Izquierdo, 2009; Warden, 1997)

tested and revealed the effectiveness of form-focused instruction (FFI) including several

noticing, awareness, and practice activities while drawing L2 learners’ attention to

sublexical cues. In particular, one noteworthy finding is that, as documented by Harley

(1998) and Lyster (2004), L2 learners tend to pronounce French articles in an ambiguous
 2

manner such as a hybrid form between un and une and between le and la. Lyster (2004)

stated that “This strategy eases the L2 learner’s burden of having to accurately mark

grammatical gender so frequently” (p. 416).

Considering that L2 learners have difficulty producing the French vowels /œ̃/ (in

un) and /y/ (in une) (e.g., Li & Rosen, 2016), however, it may be the case that French L2

learners misarticulate the French articles ambiguously not only to ease the cognitive

burden of having to assign grammatical gender, but also because they have difficulty

perceiving and producing the sounds per se (i.e., /œ̃/ and /yn/). Accordingly, it is

hypothesized that, in order for L2 learners to demonstrate targetlike performance

regarding French grammatical gender, FFI targeting pronunciation of the sounds as well

as sublexical cues should be implemented in classroom instruction.

This dissertation entails a quasi-experimental study in French L2 classrooms with

the following instructional conditions: (a) FFI on only sublexical cues, (b) FFI on both

sublexical cues and pronunciation, and (c) control condition. The instructional

techniques, drawing L2 learners’ attention to sublexical cues, are implemented in the first

FFI condition. Pronunciation instruction increasing L2 learners’ production and

perception accuracy of the sounds un, une, le, and la is added to the FFI on sublexical

cues for the second FFI condition.

By administering various tasks, the current study is expected to provide empirical

evidence as to how L2 learners can overcome their difficulty in producing grammatical

gender in a targetlike manner. Based on the prediction that phonological inability might

be one of the factors resulting in L2 learners’ difficulty, the current study is expected to
 3

demonstrate the benefits of FFI targeting both sublexical and phonological information in

the L2 acquisition of French grammatical gender.

In addition, the present study focuses on L2 learners’ individual differences.

Given that L2 learners’ executive function (EF) skills are important catalysts in L2

learning (Darcy, Mora, & Daidone, 2016; Kapa & Colombo, 2014; Linck, Osthus, Koeth,

& Bunting, 2014), the present study attempts to examine whether the extent to which L2

learners benefit from the FFI conditions is mediated by their EF skills—inhibitory

control, nonverbal visuospatial working memory, and cognitive flexibility.

The current study is expected to have many implications for L2 acquisition,

education, and psycholinguistics. By showing that L2 phonological ability plays an

important role in the L2 acquisition of French grammatical gender, the study aims to shed

light on the importance of L2 pronunciation instruction and its interdependence with

lexical and morphological domains. In addition, the present study will provide L2

practitioners with research-based instructional techniques for L2 pronunciation

instruction, which they can easily adapt to L2 instruction including grammatical targets

such as French grammatical gender. Moreover, based on the results regarding L2

learners’ EF skills, the present study will also bring to the fore the roles of L2 learners’

EF skills in the realm of L2 acquisition and psycholinguistics.

1. 1. Structure of the Dissertation

This dissertation consists of a total of six chapters. Following this introductory

chapter, Chapter 2 presents the background to the present study, beginning with the

motivations of the study. After a brief introduction to French grammatical gender, the
 4

importance of noun endings as indicators of French grammatical gender attribution is

introduced. Previous FFI studies targeting French grammatical gender are also

summarized. With respect to EF skills, previous studies focusing on inhibitory control,

working memory, and cognitive flexibility are presented. Finally, the research questions

of the current study are presented along with possible implications in the field of L2

acquisition, education, and psycholinguistics.

Chapter 3 entails the methodology of the present study. Following the description

of the participants, the procedure of the study and target noun endings are introduced. All

instructional conditions and measures are then described along with the procedure of data

preparation for data analysis.

Chapter 4 includes the results of the current study. This chapter presents

descriptive and inferential statistics with respect to the following three sub-sections: (a)

the effects of two FFI conditions, (b) the variables affecting the performance of the

picture-description task, and (c) the roles of EF skills in learning gains.

Chapter 5 discusses the results, highlighting the differential effects of the two FFI

conditions on the L2 acquisition of French grammatical gender. Based on the results, the

chapter addresses the importance of L2 phonological knowledge and pronunciation

instruction in the L2 acquisition of French grammatical gender. It also states the roles of

EF skills in L2 instruction by revealing that the extent to which L2 learners benefit from

FFI is mediated by their EF skills.

Lastly, Chapter 6 concludes this dissertation by summarizing the present study.

After elaborating the pedagogical implications of the current study, this chapter proposes

future directions drawing on the limitations of the present study.

 5

Chapter 2

Background

This chapter provides the background to the current study. The chapter begins by

illustrating the motivations of the present study, followed by a literature review focusing

on French grammatical gender, FFI, and EF skills. After a brief introduction to French

grammatical gender, the importance of noun endings as indicators of French grammatical

gender attribution is presented. The pedagogical benefits of FFI in L2 learning are

highlighted along with the effectiveness of FFI on the L2 acquisition of French

grammatical gender. Considering that the extent to which L2 learners benefit from FFI

might be mediated by their individual EF skills, the roles of EF skills—inhibitory control,

working memory, and cognitive flexibility—in L2 learning are then introduced. The

chapter concludes by presenting the research questions with possible implications.

2. 1. Motivation of the Present Study

Harley (1998) contended that French grammatical gender is not something that

L2 learners can acquire incidentally from language exposure. In light of the importance

of noun endings as predictors of gender attribution (Tucker et al., 1977; Tucker et al.,

1968), previous studies (Harley, 1998, Lyster, 2004; Lyster & Izquierdo, 2009; Warden,

1997) implemented FFI drawing L2 learners’ attention to noun endings as predictors of

grammatical gender attribution and confirmed its pedagogical effectiveness. One

noteworthy finding from previous studies is that French L2 learners tended to develop an

interlanguage strategy resulting in ambiguous pronunciation of French articles and thus a

 6

lack of intelligibility and grammatical inaccuracy after all. For instance, Lyster (2004)

noted that

In an earlier pilot study (…) we were left with a considerable amount of

untranscribable data. As documented by Harley (1998), immersion

students have developed a strategy of using hybrid forms that sound like a

combination of both un and une or (…) Data collected during the piloting

caused hours of discussion among members of the research team as we

debated whether students said un or une or a hybrid form (p. 416).

Lyster (2004) stated that “This strategy eases the L2 learner’s burden of having to

accurately mark grammatical gender so frequently” (p. 416).

According to Li and Rosen (2016), French L2 learners (French immersion

students in Southern Alberta) have difficulty pronouncing the French nasal vowel /œ̃/ (in

un) and the oral vowel /y/ (in une). In particular, the nasal vowel /œ̃/ was the most

challenging for the L2 learners and was not produced in a targetlike manner even after

intensive L2 learning. Presumably, it may therefore be the case that French L2 learners

misarticulate the French articles ambiguously not because they lack knowledge of

grammatical gender. Rather, they might have difficulty perceiving and producing the

sounds per se of the French articles (particularly, un and une), which might in turn

predispose them to produce ambiguous pronunciation. Some studies (Goad & White,

2006; Goad, White, & Steele, 2003) also speculated that L2 morphosyntactic errors might

be related to the properties of first language (L1) phonology. In this regard, the current

study predicts that L2 learners’ phonological knowledge is one of the variables affecting
 7

the L2 acquisition of French grammatical gender and that pronunciation instruction on

the sounds of French articles facilitates the L2 acquisition of French grammatical gender.

The effects of pronunciation instruction have been investigated in classroom

settings focusing on multiple L2 segmental and suprasegmental targets (e.g., Lee &

Lyster, 2016a; Saito, 2013; Saito & Lyster, 2012; Saito & Wu, 2014). In particular, Saito

(2013) emphasized the role of explicit phonetic information (e.g., articulatory gestures) in

pronunciation instruction, arguing that explicit phonetic information might facilitate

phonetically driven L2 pronunciation learning. In this sense, by implementing

pronunciation instruction including explicit phonetic information, the present study

hypothesizes that pronunciation instruction targeting the sounds of French articles will

facilitate the L2 acquisition of French grammatical gender by enabling L2 learners to

develop more targetlike pronunciation of the sounds of French articles.

Finally, previous studies (Darcy et al., 2016; Kapa & Colombo, 2014; Linck et al.,

2014) showed that L2 learners’ EF skills are important predictors of L2 learning. In light

of such findings, the present study hypothesizes that EF skills might be significant

predictors of the extent to which L2 students benefit from FFI. Based on the motivations

of the present study, I now introduce previous studies regarding French grammatical

gender, FFI, and EF skills.

2. 2. French Grammatical Gender

Although natural gender exists in all animals, the way in which natural gender is

expressed in language is language-specific. For instance, while gender in English is

assigned to some lexical items (e.g., a husband, a wife, a son, a daughter) and some
 8

pronouns (e.g., he, she, him, her), English is generally known as a gender-neutral

language. In contrast, in some languages such as French, gender is assigned to all animate

and inanimate nouns (i.e., gender attribution), and gender agreement is required within

noun phrases comprising a determiner, a noun, and an adjective as in (1):

(1)

a. mon nouveau livre

b. ma nouvelle maison

According to Sera et al. (2002), French entails two gender categories (i.e.,

masculine and feminine). It is not a case-based system; that is, the determiner mon is

immutable regardless of whether it is a determiner for the object noun or for the subject

noun as long as the noun is masculine as in (2):

(2)

a. J’aime mon ami.

b. Mon ami m’aime.

In French, grammatical gender, as opposed to biological gender, connotes the

gender of generic nouns such as inanimate nouns and some animate nouns without

considering a semantic basis for gender attribution (Sokolik & Smith, 1992). For

instance, un chapeau is masculine, whereas une casquette is feminine; un calmar is

masculine, but une crevette is feminine. For nouns related to humans and certain animals,

however, their gender attribution is usually determined by their biological gender. For

instance, un garçon and un lion are masculine, and une fille and une lionne are feminine.

From generative perspectives, gender is an interpretable feature of French nouns,

whereas it is an uninterpretable (formal) feature of French determiners and adjectives.

 9

Therefore, the latter components should be checked through agreement in the syntax

(Chomsky, 1995). More specifically, Bernstein (1993), Picallo (1991), and Ritter (1991)

stipulated that gender and number are functional categories in the DP (determiner phrase)

above the NP (noun phrase).

Carstens (2000) argued that a noun is embedded in the syntactic tree as a head of

a noun phrase and entails an interpretable gender feature. The noun is raised to the AgrP

(agreement phrase) and then to D (for a determiner) in which it checks uninterpretable

gender features in specifier-head (for noun-adjective concord) and head-head (for

determiner-noun concord) relations. Gender is lexically assigned to nouns in French, and

thus gender agreement is a syntactic feature-checking operation by the syntax (Paradis &

Prévost, 2004).

The most prominent psycholinguistic model presupposes that grammatical gender

is encoded as a property of nouns at a representational level different from those

specifying the corresponding conceptual and phonological information (Cubelli, Lotto,

Paolieri, Girelli, & Job, 2005). The Word-Form Encoding by Activation and Verification

Model (WEAVER++), which was proposed by Roelofs (1992) and revised by Levelt,

Roelofs, and Meyer (1999), postulates three main layers in a word: The top layer conveys

the meaning of the word with the aid of a network of conceptual connections. The

intermediate layer includes the abstract lexical representation (lemma), which is related to

nodes concerning the syntactic properties of the word such as grammatical gender. The

third layer specifies the phonological form (lexeme) of the word. According to the model,

the phonological form of the word is only activated after its lemma is selected, which

might in turn be activated by its relevant conceptual node. In addition, gender

 10

information is only available in the syntactic environment through competition at the

level of gender feature selection, preceding the access to the phonological form.

The alternative model, the Independent Network Model (IN; Caramazza, 1997;

Caramazza & Miozzo, 1997), hypothesizes three separate networks; that is, lexical-

semantic, syntactic, and phonological information. In contrast to the WEAVER++

(Levelt et al., 1999; Roelofs, 1992), this model proposes that semantic representations

can activate word forms directly, without assuming an intervening lemma node. The

syntactic features of a word thus require the prior selection of the semantically and

syntactically specified, modality-specific lexical forms. Therefore, nominal gender does

not interfere with the selection of a phonological representation of a word, and lexical

selection is only supplied by semantic information without assessing the syntactic

features. The model thus presupposes that, instead of competition at the level of gender

feature selection, gender information is an automatic sequence as a result of the selection

of the modality-specific lexical forms.

Overall, both models posit that gender is only selected in gender-marked

utterances. While hypothesizing that grammatical gender emerges at the phrase level, the

models theoretically predict that grammatical gender might be overlooked in the

production of bare nouns. Empirically, according to La Heij, Mark, Sander, and

Willeboordse (1998), when target and distractor nouns had the same grammatical gender,

participants showed shorter naming latencies when asked to produce noun phrases.

However, such a gender congruity effect was not found in the production of bare nouns.

Cubelli et al. (2005) found that bare noun production times were slower when target and

distractor nouns possessed the same grammatical gender than when they had different
 11

grammatical gender. They thus argued that, in contrast to the WEAVER++ (Levelt et al.,

1999; Roelofs, 1992), the selection of grammatical gender is mandatory (i.e., even

outside a sentential context). Moreover, in contrast to the IN (Caramazza, 1997;

Caramazza & Miozzo, 1997), the selection of grammatical gender is not automatic, but it

entails a competitive process preceding the access to morpho-phonological forms.

These findings in Cubelli et al. (2005) are also compatible with those in Tucker et

al. (1977). In their study, when French L1 speakers were asked to determine the

grammatical gender of rare nouns and pseudonouns, the French L1 speakers reported that

they tested each noun with masculine and feminine indefinite articles respectively (i.e.,

grammatical gender emerges at the phrase level) and then decided which one sounded

better (i.e., competitive process). One of the interesting findings is that they tended to rely

on noun endings unconsciously to determine which one sounded better (i.e., during the

competitive process). In what follows, the importance of noun endings as indicators of

French grammatical gender attribution is introduced.

2. 3. Noun Endings as Predictors of French Grammatical Gender

It is known that French L1 speakers master the grammatical gender system by the

age of 3 (Clark, 1985; Karmiloff-Smith, 1979; van Heugten & Shi, 2009). Tucker et al.

(1968) found that French L1 speakers assign the grammatical gender of pseudonouns

based on their noun endings. For example, for the pseudonouns florillon and florateur,

French L1 speakers categorized them as masculine nouns owing to the masculine markers

‘-illon’ and ‘-eur’. For the pseudonoun feuillation, they considered it a feminine noun due

to the feminine marker ‘-tion’.

 12

According to Tucker et al. (1977), French L1 speakers develop a very powerful

and implicit grammatical gender system without any explicit instruction. The study found

that French L1 speakers (ages 7-17) exploit noun endings to predict the gender attribution

of rare nouns and pseudonouns. French L1 speakers “focus on the ending as the most

probable gender marker, and then scan backwards into the words until they can determine

in which particular subcontext the terminal phone occurs” (p. 62). For instance, for the

noun nation, French L1 speakers scan its noun endings from ‘-on’ (masculine) to ‘-tion’

(feminine) and then assign its grammatical gender correctly, une nation. Therefore, noun

endings “co-occur in a systematic and predictable manner” with gender attribution (p.

57). In a similar vein, Karmiloff-Smith (1979) also found that L1 speakers develop an

accurate grammatical system clearly based on suffixes.

A corpus analysis by Lyster (2006) supports the role of noun endings as

predictors of gender attribution, revealing that French grammatical gender is rule-

governed, having noun endings that predict gender attribution in a systematic manner.

The study analyzed a corpus of 9,991 nouns appearing in Le Robert Junior Illustré. He

found that gender attribution is vastly predicted by noun endings operationalized as

orthographic representations of rhymes including either a nucleus for vocalic endings or a

nucleus-and-coda blend for consonantal endings. For instance, more than 90% of nouns

ending with ‘-ent’, ‘-ant’, ‘-eau’, and ‘-ai’ are masculine, whereas more than 90% of

nouns ending with ‘-che’, ‘-esse’, and ‘-asse’ are feminine.

The way in which noun endings are operationalized seems partly phonological.

For instance, the final phonemes /ɑ̃/ and /o/ are mostly masculine, while the final

phonemes /z/ and /ʃ / are mostly feminine (more than 90% in Lyster, 2006). However,
 13

there are multiple instances in which final phonemes interact with orthographic

representations, which override phonological representations in predicting gender

attribution. For example, the final phonemes of the noun endings ‘-asse’, ‘-isse’, and

‘-esse’ are the same as those of the noun endings ‘-as’, ‘-is’, and ‘-ès’. Yet, the former

endings are mostly feminine (93%), whereas the latter endings are mostly masculine

(99%). Accordingly, the orthographic representations of the noun endings found in Lyster

(2006), some of which were employed in the current study, are more reliable than their

final phonemes.

In this regard, French L1 speakers develop an accurate grammatical gender

system at an early age and employ noun endings as reliable predictors of gender

attribution (see also Desrochers, Paivio, & Desrochers, 1989; Holmes & de la Bâtie;

1999). On the other hand, L2 learners commonly have a great deal of difficulty acquiring

grammatical gender in a targetlike manner. For instance, despite a number of years in

French immersion education, 11- to 12-year-old immersion students seldom

discriminated feminine nouns from masculine nouns while overgeneralizing either

masculine or feminine gender markers (Harley, 1979). Bartning (2000) also found that

even advanced L2 learners have difficulty with French grammatical gender, particularly

with indefinite articles. A number of previous studies (e.g., Harley, 1998; Lyster, 2004)

reported similar findings in which the researchers contended that French grammatical

gender is still problematic in spite of the high frequency of gender markings in the input.

Surridge and Lessard (1984) conducted a study to investigate how L2 learners

determine gender attribution. The study showed that L2 learners are sensitive to

morphological cues to determine gender attribution in a similar manner to French L1

 14

speakers. Hardison (1992) also reported that L2 learners tended to focus on noun endings

(e.g., both phonemic and orthographic representations) to predict gender attribution.

Carroll (1989) argued that French L1 speakers acquire and process determiners

and nouns as co-indexed chunks. For example, for the noun passeport, they encode and

retrieve the noun with its determiner as a chunk such as /ləpɑspɔʀ/, lepasseport. When

the chunk is analyzed as separate constituents, the noun still encodes its inherent gender

information. However, L2 learners are more likely to encode and retrieve them as

separate entities. Therefore, Carroll (1989) proposed that L2 learners need mnemonic

strategies and rules, which “could provide the advanced learner not only with a

reasonably accurate system but also with a mechanism for guessing the gender of new

items” (p. 580). In particular, the mnemonic strategies and rules could be related to

morphological aspects, sensitive to the suffixes of target nouns.

Given that L2 learners can benefit from morphological information embedded in

word-internal properties (e.g., sublexical cues) to predict gender attribution, L2

researchers have implemented instructional techniques drawing L2 learners’ attention to

the morphological information and tested their pedagogical effectiveness by adopting

several FFI techniques in classroom settings. In what follows, the definition and

instructional components of FFI are presented along with its effectiveness on the L2

acquisition of French grammatical gender.

 15

2. 4. Form-Focused Instruction on Second Language Learning

2. 4. 1. Definition and Components of Form-Focused Instruction

Spada (1997) defined FFI as “any pedagogical effort which is used to draw the

learners’ attention to form either implicitly or explicitly... within meaning-based

approaches to L2 instruction [and] in which a focus on language is provided in either

spontaneous or predetermined ways” (p. 73). According to Ranta and Lyster (2018), FFI

consists of proactive and reactive FFI techniques (see Figure 1). For instance, proactive

FFI begins with input enhancement which induces L2 learners to notice and process

linguistic targets in the input, after which metalinguistic explanations are provided.

Proactive FFI concludes with controlled and spontaneous practice to facilitate automatic

and fluent use of the linguistic targets. Reactive FFI is offered as corrective feedback

responding to L2 learners’ erroneous utterances so they can restructure their

interlanguage system.

Figure 1. Components of proactive form-focused instruction (Ranta & Lyster, 2018,

p. 43)

Its implementation is supported by several L2 theories. For example, the noticing

hypothesis (Schmidt, 2001) proposed that noticing L2 linguistic targets is necessary and

offers an initial scaffolding for successful L2 acquisition. Input enhancement in FFI

 16

serves as an instructional tool to incite L2 learners’ noticing. In addition, according to

skill acquisition theory (DeKeyser, 1998, 2001; Lyster & Sato, 2013), there are two types

of L2 knowledge: declarative and procedural knowledge. Declarative knowledge includes

metalinguistic information such as grammatical rules, while procedural knowledge entails

abilities to apply the metalinguistic information during actual use of L2. Therefore,

metalinguistic explanations in FFI support the development of declarative knowledge,

and practice opportunities enable learners to proceduralize the declarative knowledge

(i.e., procedural knowledge). Moreover, practice opportunities are also compatible with

the output hypothesis (Swain, 1985, 1995), which posits the importance of language

output in L2 learning. The effectiveness of FFI has been tested and confirmed in various

instructional settings while focusing on several linguistic targets (Gooch, Saito, & Lyster,

2016; Laufer & Girsai, 2008; Lee & Lyster, 2016a; Lyster, 2004; Nguyen, Pham, &

Pham, 2012; Spada, Jessop, Tomita, Suzuki, & Valeo, 2014). Next, I introduce FFI

studies targeting French grammatical gender.

2. 4. 2. Effectiveness of Form-Focused Instruction on French Grammatical Gender

Most French grammarians argue that French grammatical gender is arbitrary and

unsystematic, particularly in the case of inanimate nouns (e.g., Laurin & Jacob, 2006).

Therefore, it has been suggested that gender attribution needs to be acquired on an item-

by-item basis. Yet, previous studies (Tucker et al., 1977; Tucker et al., 1968) found that

gender attribution is highly related to noun endings. As such, it was suggested that L2

learners need to be aware of noun endings in order to develop skills to assign gender

accurately and that instructional techniques which draw learners’ attention to noun

endings are worth considering in classroom settings. Four FFI studies (Harley, 1998;
 17

Lyster, 2004; Lyster & Izquierdo, 2009; Warden 1997) targeting French grammatical

gender were conducted in which several pedagogical techniques inciting L2 learners to

notice and internalize noun endings as predictors of grammatical gender were developed

and empirically tested. Table 1 summarizes the studies.

Table 1

Form-Focused Instruction on French Grammatical Gender

FFI
Study Participants Groups Measures
hours
Pretest + Two posttests
Grade 11
● Listening
French 12 hours
Warden 1. FFI ● Written endings
immersion over
(1997) 2. Control ● Agreement
students 5 weeks
● Writing attribution
(n = 62)
● Oral production
Posttests:
● Listening: FFI > Control
● Written endings: FFI > Control
Results
● Agreement: FFI > Control
● Writing attribution: No significant differences
● Oral production: No significant differences
Grade 2
Pretest + Two posttests
French 20 mins
● Aural discrimination
Harley immersion daily 1. FFI
● Binary-choice (le or la)
(1998) students over 2. Control
● Picture-description
(n = 210 5 weeks
● Object-identification
~ 300)
Between-group contrasts (Posttests):
● Aural discrimination: FFI > Control
● Binary-choice (le or la): FFI > Control
● Picture-description: FFI > Control
● Object-identification : No significant differences
Results
Within-group contrasts (FFI group):
● Aural discrimination: Posttests > Pretest
● Binary-choice (le or la): Posttests > Pretest
● Picture-description: Posttests > Pretest
● Object-identification : No significant differences
 18

Grade 5 Pretest + Two posttests

1. FFI+Recast
Frech 9 hours ● Binary-choice
Lyster 2. FFI+Prompt
immersion over ● Text-completion
(2004) 3. FFI-only
students 5 weeks ● Object-identification
4. Control
(n = 179) ● Picture-description
Posttest 1:
● Binary-choice: FFI+P > FFI+R, FFI, Control (FFI+R > Control)
● Text-completion: FFI+P > FFI+R, FFI, Control (FFI+R, FFI > Control)
● Object-identification: FFI+P > Control
● Picture-description: FFI+P > Control
Results
Posttest 2:
● Binary-choice: FFI+P > FFI+R, FFI, Control (FFI+R, FFI > Control)
● Text-completion: FFI+P > FFI+R, FFI, Control
● Object-identification: FFI+P, FFI+R, FFI > Control
● Picture-description: FFI+P, FFI+R, FFI > Control
Adult L2 Pretest + Two posttests
Lyster & 3 hours
learners of 1. FFI+Recast ● Reaction-time binary-choice
Izquierdo over
French 2. FFI+Prompt ● Object-identification
(2009) 2 weeks
(n = 25) ● Picture-description
Results Both groups showed significant improvement over time.

As shown in Table 1, all studies revealed the effectiveness of FFI on increasing

L2 learners’ accuracy by drawing their attention to noun endings as predictors of gender

attribution. One of the interesting reports from previous studies (Harley, 1998; Lyster,

2004) is that L2 learners tended to produce ambiguous pronunciation resembling a hybrid

form between un and une and between le and la, resulting in a lack of both intelligibility

and grammatical inaccuracy. In a similar vein, previous studies showed mixed effects in

oral production. For instance, Lyster (2004) and Lyster and Izquierdo (2009) found

effects for FFI in the two oral tasks (i.e., object-identification and picture-description

tasks). However, Warden (1997) did not find any significant effects in the oral production

task, and neither did Harley (1998) in the controlled oral production task (i.e., object-

identification task).
 19

Considering that /œ̃/ (in un) and /y/ (in une) are difficult for L2 learners to acquire

(e.g., Li & Rosen, 2016), L2 learners might have had difficulty articulating the sounds of

French articles in spite of their targetlike grammatical accuracy in gender attribution. To

provide empirical evidence regarding this hypothesis, the present study aims to examine

whether L2 learners’ phonological knowledge impedes them from showing targetlike

performance in oral production, and if so, whether FFI on pronunciation facilitates the L2

acquisition of French grammatical gender.

2. 5. Executive Functions Skills in Second Language Learning

According to Diamond (2012), EF skills “are a family of control functions needed

when you have to concentrate and think, when acting on your initial impulse might be ill-

advised. These functions depend on a neural circuit in which the prefrontal cortex plays a

prominent role” (p. 336). There is a general consensus that there are three key

components in EF skills: (a) inhibitory control, (b) working memory, and (c) cognitive

flexibility. In what follows, I introduce each component focusing on its role in L2

acquisition.

2. 5. 1. Inhibitory Control

Inhibitory control is a cognitive process that inhibits behavior responses to stimuli

(Kok, 1999; Nigg, 2000) or that suppresses predominant responses in a deliberate and

controlled manner (Miyake et al., 2000). In L2 acquisition, inhibitory control enables L2

learners to process and produce an L2 while inhibiting their L1 (Green, 1998). In

particular, inhibitory control is proportional to the degree of the activation of the

 20

representations to be suppressed; higher inhibitory control is thus required when

processing and producing an L2 in contrast to an L1 (Costa & Santesteban, 2004).

Empirically, Darcy et al. (2016) found that L2 learners with high inhibitory

control were likely to show more targetlike speech perception and production accuracy of

L2 segments. In addition, the relationship between inhibitory control and perception was

stronger than the relationship between inhibitory control and production. Darcy et al.

(2016) argued that high inhibitory control of L1 allows L2 learners to facilitate “the

processing of phonologically relevant acoustic information in the L2 input, which in turn

might lead to more accurate L2 phonological representations” (p. 742). Kapa and

Colombo (2014) also tried to tease apart the role of inhibitory control in L2 learning

using an artificial language paradigm. They found that adults’ inhibitory control was a

significant predictor of L2 learning after controlling for L1 vocabulary size and working

memory.

With respect to the role of inhibitory control, most studies were conducted

targeting bilingual speakers (e.g., Martin-Rhee & Bialystok, 2008; Poarch & van Hell,

2012), with a few studies targeting instructed L2 learners. The studies correlated L2

learners’ current L2 knowledge and their inhibitory control, concluding that high L2

accuracy results from high inhibitory control. Given the importance of exploring the role

of inhibitory control in the learning process, the current study investigates whether

learning gains are mediated by L2 learners’ inhibitory control. Based on Darcy et al.

(2016), L2 students with high inhibitory control are expected to increase their accuracy of

perceiving and producing the sounds of French articles to a greater degree than those with

low inhibitory control.

 21

2. 5. 2. Working Memory

Working memory is a cognitive system that is responsible for information

holding, temporal processing, and maintenance (Miyake & Shah, 1999). Baddely and

Hitch’s (1974) multicomponent model of working memory consists of the central

executive, the phonological loop, and the visuospatial sketchpad. The central executive is

responsible for attention control while regulating the integration of information and

supervising two slave systems (i.e., the phonological loop and the visuospatial

sketchpad). The phonological loop is responsible for processing and storing phonological

information (e.g., verbal and acoustic information), whereas the visuospatial sketchpad is

responsible for processing and storing visual, spatial, and kinesthetic information

(Baddeley, 2003). Baddeley (2000) added another component in the model—the episodic

buffer—which controls the link between working memory and long-term memory.

Working memory is found to be an important predictor of the learning of various

L2 linguistic targets. Research has found that, compared to learners with low working

memory, learners with high working memory exhibit advantages in their rate of L2

vocabulary learning (Atkins & Baddeley, 1998) and also in having to look up fewer

words to understand a story (Chun & Payne, 2004). Other studies have shown benefits for

learners with high working memory in L2 grammar learning (French & O’Brien, 2008),

online L2 parsing performance (Juffs, 2004), structural priming in L2 speech production

(McDonough & Kim, 2016), and L2 morphosyntactic pattern learning (McDonough &

Trofimovich, 2016). The importance of working memory has also been highlighted by

several meta-analyses. For instance, Linck et al. (2014) conducted a meta-analysis to

investigate the impact of working memory in L2 comprehension and production, which

 22

found a robust and positive relationship between L2 learners’ working memory and L2

learning outcomes. In particular, their analysis revealed larger effect sizes for the

executive control in contrast to storage components and for verbal rather than nonverbal

working memory.

Most of this research, however, has investigated verbal working memory, mainly

tapping into the phonological loop. To move this line of research forward, the present

study investigates the role of nonverbal visuospatial working memory for the following

two reasons. First, some studies (e.g., Gangopadhyay, Davidson, Weismer, &

Kaushanskaya, 2016) stated that any correlations between verbal working memory and

L2 linguistic performance “may have been due to an overlap in language, rather than

WM (working memory) demands, between language processing and WM tasks” (p. 188).

Second, FFI in the current study was designed to draw L2 learners’ attention to noun

endings as predictors of grammatical gender. Therefore, L2 learners receiving FFI will be

pushed to focus on noun endings to determine grammatical gender (i.e., orthographic

processing focusing on sublexical cues). Orthographic processing induces learners to

detect “the formation of visual representations of letters, letter patterns, and sequences of

letters that serve to map spatially the temporal sequence of phonemes within words”

(Pham & Hasson, 2014, p. 474). Therefore, there is a possibility that L2 learners’

visuospatial working memory will be an important predictor of the degree to which L2

learners notice and internalize noun endings as predictors of grammatical gender during

FFI sessions.
 23

2. 5. 3. Cognitive Flexibility

Cognitive flexibility involves the ability to switch between two different concepts

or mental sets (Miyake et al., 2000). Most previous studies investigating cognitive

flexibility compared bilingual and monolingual speakers. For instance, Bialystok and

Viswanathan (2009) found that bilinguals were faster than monolinguals in task

switching while performing the Trail-Making Test. Similarly, Nicolay and Poncelet

(2013) showed that students in immersion programs were significantly faster than

monolinguals in switching attention during the Test for Attentional Performance in

Children. In light of the findings, it was presupposed that small switching costs are due to

a bilingual advantage (i.e., frequent switching between two languages). L2 learners with

high cognitive flexibility are likely to show more rapid and accurate switching between

their L1 and L2 (see also Seçer, 2016).

There are mixed results regarding the role of cognitive flexibility in L2

acquisition. For instance, Kapa and Colombo (2014) demonstrated that children’s L2

performance in learning an artificial language was predicted by their cognitive flexibility.

On the other hand, Stone and Pili-Moss (2015) failed to find any significant relationship

between cognitive flexibility and the L2 acquisition of morphosyntax in Brocanto2, an

artificial language.

In this regard, the current study attempts to test whether the extent to which L2

learners benefit from FFI is mediated by their cognitive flexibility. In the present study,

none of the students’ L1s have grammatical gender. It predicts that those having high

cognitive flexibility will be at an advantage in becoming aware of grammatical gender,

 24

which is a missing feature in their L1s, and showing targetlike L2 performance by having

rapid and accurate switching between their gender-neutral L1s and L2 French.

2. 6. Research Questions

In light of previous research studies and the motivations of the current study, the

research questions are as follows:

1. To improve their accuracy in French grammatical gender, do French L2

learners benefit more from FFI on only sublexical cues or from FFI on

both sublexical cues and pronunciation?

2. To what extent are the benefits of these FFI conditions mediated by

individual differences in L2 learners’ EF skills such as inhibitory control,

visuospatial working memory, and cognitive flexibility?

According to Grüter, Lew-Williams, and Fernald (2012), there are three primary sources

of difficulty hindering L2 learners in their L2 acquisition of grammatical gender: (a)

difficulty at the level of gender attribution (lexical knowledge); (b) difficulty at the level

of gender agreement (syntactic knowledge); and (c) difficulty with assessing and

deploying the lexical and syntactic knowledge in online production. In particualr, they

found that L2 learners’ gender attribution errors were more than 10 times as frequent as

their gender agreement errors. They thus concluded that nontargetlike L2 lexical—rather

than syntactic—representations might be a primary factor which prevents L2 learners

from acquiring grammatical gender in a targetlike manner. Accordingly, the current study

focuses on the L2 acquisition of gender attribution (not gender agreement) in French.

 25

The present study is expected to have many implications for L2 acquisition,

education, and psycholinguistics. By showing that L2 phonological ability plays an

important role in the L2 acquisition of French grammatical gender, the study sheds light

on the importance of L2 pronunciation instruction and its interdependence with lexical

and morphological domains. In addition, the current study provides L2 practitioners with

research-based instructional techniques for L2 pronunciation instruction, which they can

adapt to L2 instruction targeting French grammatical gender. Finally, the study is

expected to highlight the roles of L2 learners’ EF skills in L2 instruction while adding

emperical evidence to the exisiting body of EF literature.

In order to answer the research questions, a quasi-experimental study was

conducted in French L2 classrooms. The next chapter introduces its research

methodology.
 26

Chapter 3

Methodology

This chapter presents the methodology of the current study. The procedure of the

study and target noun endings are provided following the description of the participants.

After an overview of the instructional sessions, a detailed description of each

instructional condition is presented. Each measure is also described along with an

overview of the measures. Finally, this chapter concludes by explaining the procedure of

data preparation for data analysis and summarizing the chapter.

3. 1. Participants

Six classrooms (Classes 1-6) in Elementary French 1 (FRSL 207/208) participated

in the current study. The course was one of the French L2 courses offered by the McGill

French Language Centre in the Faculty of Arts in Fall 2017. The course was open to

French L2 learners who had completed 100-level French L2 courses, but not taken Grade

12- or 13-level French L2 courses in Canada or any French-speaking countries. Initially,

a total of 162 students in the six classrooms participated in the present study. However, in

order to control for any L1 effects on the acquisition of French grammatical gender,

students whose L1 had grammatical gender (e.g., Spanish and Portuguese) were removed

from the analyses. As a result, a total of 140 students participated in the present study.

Table 2 summarizes their background.

 27

Table 2

Background of Participating Students

Mean age
Mean age
(when learning
Class (when Sex L1 background
French for the
participating)
first time)
English (n = 9)
Female
Turkish (n = 3)
Class 1 19.1 (n = 12) 14.8
Japanese (n = 2)
(n = 17) (SD = 1.20) Male (SD = 5.11)
Mandarin (n = 2)
(n = 5)
Bengali (n = 1)
Mandarin (n = 11)
Female
English (n = 10)
Class 2 21.6 (n = 21) 15.8
Japanese (n = 1)
(n = 24) (SD = 5.05) Male (SD = 5.61)
Korean (n = 1)
(n = 3)
Persian (n = 1)
English (n = 13)
Female
Mandarin (n = 9)
Class 3 19.8 (n = 13) 15.1
Bengali (n = 1)
(n = 25) (SD = 1.66) Male (SD = 5.43)
Japanese (n = 1)
(n = 12)
Turkish (n = 1)
English (n = 16)
Female
Mandarin (n = 5)
Class 4 19.8 (n = 17) 15.1
Japanese (n = 1)
(n = 24) (SD = 2.43) Male (SD = 4.37)
Turkish (n = 1)
(n = 7)
Vietnamese (n = 1)
Mandarin (n = 10)
Female English (n = 8)
Class 5 21.6 (n = 15) Korean (n = 2) 16.1
(n = 23) (SD = 3.23) Male Indonesian (n = 1) (SD = 5.85)
(n = 8) Japanese (n = 1)
Persian (n = 1)
Female English (n = 12)
Class 6 20.5 (n = 19) Mandarin (n = 12) 17.4
(n = 27) (SD = 3.52) Male Vietnamese (n = 2) (SD = 5.91)
(n = 8) Turkish (n = 1)

None of their L1s have /œ̃/ (in un) as a separate phoneme. Mandarin and Turkish

have /y/ (in une), whereas the other L1s do not. None of the participants had French-

speaking parents nor lived in any French-speaking countries before coming to Montreal
 28

for their post-secondary education. All participants reported that they seldom spoke

French outside of their classrooms and began to learn French as their L2 in instructional

settings (mostly, at their secondary school or McGill University). Overall, the participants

stated that they were good at learning new languages (M = 5.24, SD = 1.24, Scale 1~7).

In addition, most of the participants had also learned various L2s in addition to French,

some of which had grammatical gender (e.g., Spanish, Italian, Hebrew, German, and

Greek).

On the first day of the course, all students who were registered in the course

completed a diagnostic test administered by the McGill French Language Centre. The

diagnostic test was mainly composed of dictation, listening, and grammar-focused tasks.

Based on the results of the test, the instructors confirmed that students’ proficiency level

was sufficient to take the course (i.e., neither too high nor too low) and that they had a

similar proficiency level across the six classes.

Five instructors (one male and four female instructors), employed as lecturers in

the McGill French Language Centre in the Faculty of Arts, also partook in the present

study. All instructors were native speakers of French with 6 to 23 years of teaching

experience. One instructor taught two separate classes (i.e., Classes 3 and 4), whereas the

other four instructors each taught one class (i.e., Classes 1, 2, 5, or 6).

A total of 38 native speakers of French—16 male and 22 female speakers with a

mean age of 23.9 (SD = 3.47)—took part in various roles in the present study: Audio-

stimuli speakers (n = 2), L1 baseline participants (n = 30), and native-speaker (NS) raters

(n = 6). Most of the native speakers of French were from Quebec and post-secondary

students at universities located in Montreal. All native speakers of French had French-
 29

speaking parents and completed their education in French (at least up to their secondary

level) while learning various L2s including English. In particular, the NS raters were

majoring in linguistics at a French-speaking university in Quebec.

All participants in the current study received monetary compensation supported

by the Social Sciences and Humanities Research Council. The two audio-stimuli

speakers, 30 L1 baseline participants, and six NS raters were paid $20, whereas the

participating students received $80. The three instructors in the two FFI conditions also

received $90 for their participation in a 1.5-hour teacher training session.

3. 2. Procedures

The two audio-stimuli speakers (one male and one female) were invited to the

research office in August 2017. They provided audio stimuli for forced-choice

identification and article-noun congruent/incongruent tasks. Before collecting the audio

stimuli, there was a 30-minute training session with a French L1-speaking research

assistant to ensure that the speakers would produce the stimuli correctly.

On the second day of the course, the researcher visited all classrooms to explain

the nature of the study, such as the prospective participants’ roles and monetary

compensation and then obtained their consent forms (see Appendix A) along with their

background information. To ensure that their participation was completely voluntary, all

instructors were asked to leave their classrooms while the researcher recruited

participants. The instructors did not know who did and did not participate in the study.

The results obtained from the current study did not affect any grades for the course. It is

also important to note that students in FRSL 207/208, regardless of their participation,
 30

received the instructional treatments designed for the present study (for those in the two

FFI conditions) and completed all tests (for all students) since those were considered part

of their course curriculum in Fall 2017. Yet, only the data from the students consenting to

partake in the study were extracted and analyzed in the current study.

Two weeks after the recruitment, the 140 French L2 participants completed a

pretest in September 2017, which consisted of the following six tasks: (a) two tasks

(grammatical judgment and text-completion tasks) to measure the accuracy of French

gender attribution; (b) two tasks (forced-choice identification and read-aloud tasks) to

measure the accuracy of perceiving and producing the sounds of French articles un, une,

le, and la; and (c) two tasks (picture-description and article-noun congruent/incongruent

tasks) to measure the accuracy of French gender attribution as well as the accuracy of

perceiving and producing the sounds of French articles. Before the instructional treatment

sessions, a 1.5-hour teacher training session was conducted by the researcher to help the

instructors implement each condition.

After the pretest, the 140 students participated in six 80-minute instructional

sessions (two sessions per week) from September to October 2017. Classes 1 and 2

received FFI on only sublexical cues (Condition 1). Classes 3 and 4 received FFI on both

sublexical cues and pronunciation (Condition 2), while classes 5 and 6 received their

regular instruction (Condition 3). During the instructional sessions, all classrooms (except

for one control classroom) were observed by research assistants. The research assistants

took field notes to document the progress of instructional sessions and liaised between

the researcher and the instructors.

 31

The field notes indicated that the instructors in the two FFI conditions

implemented the instructional treatments while following the lesson schedules and

administering instructional components in the way they were instructed during the

teacher training session. With respect to the control classroom observed by a research

assistant, the field notes revealed that the instructor did not use any instructional

components adopted in the two FFI conditions, but followed the regular curriculum.

Although the other control classroom was not observed, the instructor verbally reported

the same as above.

There was an immediate posttest on the day following the last instructional

session, and a delayed posttest was administrated approximately six weeks later. Both

posttests were composed of the pretesting tasks, and were thus administered in the same

manner as the pretest.

In order to investigate the extent to which learning gains were mediated by

students’ EF skills (i.e., inhibitory control, nonverbal visuospatial working memory, and

cognitive flexibility), the Simon Test (Simon & Rudell, 1967), the Corsi Block-Tapping

Test (Corsi, 1972), and the Wisconsin Card Sorting Test (Grant & Berg, 1948) were

administered at each of the three testing times.

At various times from October to November 2017, the 30 L1 baseline participants

completed a baseline test consisting of the pretesting tasks. The purpose of having the L1

baseline participants was twofold: (a) to ensure that the six tasks, measuring the accuracy

of French gender attribution and the accuracy of perceiving and producing the sounds of

French article, assessed linguistic skills other than any other problem-solving skills (thus,

maximum scores were expected from the L1 baseline participants) and (b) to ensure that
 32

the speech ratings from the six NS raters were reliable enough to answer the research

questions (thus, maximum scores attained by the L1 baseline participants).

All tasks were programmed by a software tool, LearningBranch

([Link] and a software package for psychological

experiments, PsyToolkit (Stoet, 2010, 2017). Therefore, all tests were administered by

means of individual computers at the Arts Multimedia Language Facility for the

participating students and at the research office for the L1 baseline participants. Each

testing session took 1 to 1.2 hours. Finally, speech samples collected from the read-aloud

and picture-description tasks were rated by the six NS raters for subsequent analyses.

3. 3. Target Noun Endings

To choose target noun endings in the current study, a corpus analysis of the

instructional materials was conducted, taking the following two factors into account: (a)

high frequency in the instructional materials and (b) more than 90% gender-predictive

values based on Lyster (2006). As a result, a total of six noun-ending types per

grammatical gender were selected as listed in Table 3. All noun endings frequently

appeared in the instructional materials while being reliable predictors of gender

attribution with more than 90% predictive values.

 33

Table 3

Summary of Target Noun Endings

Gender Type Endings Tokens

-ant (e.g., croissant)
1 43
-ent (e.g., bâtiment)
-eau (e.g., bureau)
2 -ot (e.g., mot) 28
-o (e.g., stylo)
-ais (e.g., français)
-ait (e.g., souhait)
3 27
Masculine nouns -et (e.g., billet)
-ès (e.g., succès)
-our (e.g., jour)
4 -oir (e.g., devoir) 15
-ort (e.g., confort)
5 -on (e.g., bâton) 12
-ain (e.g., refrain)
6 8
-in (e.g., matin)
-tion (e.g., question)
1 38
-sion (e.g., télévision)
2 -ie (e.g., stratégie) 31
-esse (e.g., promesse)
3 -isse (e.g., saucisse) 15
Feminine nouns -asse (e.g., classe)
-ance (e.g., chance)
4 15
-ence (e.g., science)
-ée (e.g., fumée)
5 12
-té (e.g., difficulté)
6 -che (e.g., marche) 12
Note. Tokens refer to the raw number of occurrences in the instructional materials.

3. 4. Instructional Sessions

3. 4. 1. Overview

To implement the instructional conditions designed for the present study, a total

of six 80-minute instructional sessions (two sessions per week) were implemented from

September to October 2017. In FRSL 207/208, the instructors had used a course pack

including various texts with reading, speaking, listening, and writing activities. In light of
 34

the course schedule, the following four texts in the course pack were selected for the

current study: Réponse au sphinx (Lissouba, 1994); Libre opinion : L’anglo de Saint

Pierre (Higgins, 2009); Les lettres chinoises : Lettre 2 (Chen, 1999); Les lettres chinoises

: Lettre 9 (Chen, 1999). Drawing on these texts, two instructional booklets (one for the

FFI condition on only sublexical cues and the other for the FFI condition on both

sublexical cues and pronunciation) were prepared by the researcher. Each booklet

included various instructional techniques and activities to draw students’ attention to the

linguistic targets.

For the FFI condition targeting only sublexical cues, the instructional booklet

included several instructional activities drawing students’ attention to target noun endings

that predict grammatical gender in French. With respect to the FFI condition targeting

both sublexical cues and pronunciation, pronunciation activities focusing on the sounds

of French articles were included in addition to the focus on sublexical cues. To equalize

instructional times between FFI on only sublexical cues and FFI on both sublexical cues

and pronunciation, the instruction in the former condition included more meaning-

focused activities (e.g., comprehension questions) than the instruction in the latter

condition.

It is also important to note that the FFI on only sublexical cues entailed no

pronunciation instruction on noun endings and that pronunciation instruction in the FFI

on both sublexical cues and pronunciation focused on the sounds of French articles, not

on the sounds of noun endings.

 35

3. 4. 2. Condition 1: Form-Focused Instruction on Only Sublexical Cues (Classes 1

and 2)

Each text in the instructional booklet was composed of (a) a content-focused part,

(b) a language-focused part, and (c) a content-plus-language part (see Appendix B for

sample materials).

The content-focused part consisted of two sub-parts: Reading and comprehension

questions. The instructors were asked to read aloud each text to students, who were asked

to follow along in the text as the instructors read. In each text, all target noun endings

were highlighted in bold as well as the article preceding the target noun (i.e., input

enhancement; Sharwood Smith, 1993). The instructors did not ask the students to pay

attention to the highlighted parts, but did stress the highlighted articles while reading. By

doing so, they simply drew the students’ attention to the highlighted parts without any

explicit explanation about grammatical gender. The instructors also added any

explanations to help the students understand the text. In the comprehension questions, the

students were asked to answer meaning-focused questions related to the text, after which

they formed groups of two to four students to share their responses.

The language-focused part was composed of three main activities adopted in

Lyster (2004): a cloze activity, a categorization activity, and a new word activity. During

the cloze activity, the students were asked to complete the text (the same one that they

read in the content-focused part) with blanks. They were required to write the correct

article in each blank. Only target nouns were preceded by a blank, and the endings of all

target nouns were highlighted. Once they completed the activity, the instructors asked

them to give their answers with explanations and, when necessary, provided them with
 36

the correct answers. In the categorization activity, the students were asked to categorize

the nouns targeted during the cloze activity. In a given table, they categorized all nouns

by noun endings and indicated the grammatical gender of each noun ending based on

what they identified. Once it was completed, the students were asked to give their

answers with explanations, after which the instructors explained the table with the correct

answers. The new word activity was designed to provide the students with an opportunity

to test whether the patterns they had discovered from the above two activities could also

be applied to new words. The students were asked to identify the grammatical gender

(i.e., masculine or feminine) of a set of new words whose noun endings appeared during

the previous activities. Once it was completed, the instructors asked the students to

provide their answers with explanations, and then shared the correct answers with the

students. In the case of any errors, the instructors induced them to recall the patterns they

had discovered (e.g., ‘-tion’ indique des noms féminins ou masculins?).

The content-plus-language part was mainly composed of a writing activity and a

presentation activity. In the writing activity, the students were asked to write a short

composition in French, which was related to the content of the text. The instructions of

the activity explicitly asked them to focus on grammatical gender and use correct articles

in their composition. Once it was completed, the students presented their compositions in

front of the class. While listening to them, the instructors were asked to pay attention to

their use of articles and chime in when any errors occurred. In the last session, the

students were given a total of 56 new words whose grammatical gender they were asked

to identify based on noun endings.

 37

Given that the current study did not aim to test the differential effects of different

types of feedback, the instructors were allowed to employ various feedback types at their

discretion. In the case of ambiguous pronunciation of French articles, however, the

instructors were asked not to model the correct pronunciation via recasting or explicit

correction in order to avoid overlap with Condition 2, which is described next.

3. 4. 3. Condition 2: Form-Focused Instruction on Both Sublexical Cues and

Pronunciation (Classes 3 and 4)

In addition to the instructional materials used for the FFI condition targeting only

sublexical cues, the instructor of Classes 3 and 4 offered FFI on the pronunciation of the

French indefinite and definite articles, un, une, le, and la (see Appendices C and D for

sample materials).

The instructor conducted a 20-minute pronunciation session in each of the six 80-

minute instructional sessions. The pronunciation sessions were designed to help the

students produce and perceive the sounds of French articles un, une, le, and la in a

targetlike manner. In particular, they consisted of articulation-based instruction,

segmental-level perception/production, sentence-level perception/production, and

spontaneous-level practice.

The articulation-based instruction was composed of explicit phonetic instruction

and two activities. Following Saito (2013), for the explicit phonetic instruction, the

students were given the phonetic characteristics of each sound. For instance, for each

sound, there was an IPA symbol as well as pictures illustrating phonetic information such

as nasalization, lip rounding, and tongue position. There were two pictures per sound (a

diagram showing nasalization, lip rounding, and tongue position; a photo taken of a
 38

native speaker of French pronouncing the sound). The instructor explained these

characteristics to the students and provided them with target exemplars by pronouncing

un, une, le, and la clearly four times each consecutively with exaggeration. The students

were simply asked to listen to the exemplars while paying attention to their articulatory

gestures such as nasalization, lip rounding, tongue position, and mouth opening. The

students then repeated after the instructor. For example, the instructor said une and then

asked them to repeat after her once. There were four repetitions for each sound.

In the first activity (notice-articulatory-gestures activity), there were photos of

native speakers of French pronouncing the sounds of un, une, le, and la. For each sound,

there were two photos taken of one male and one female (i.e., eight photos). The students

were asked to choose which sound each photo illustrated by recalling what they learned

during the explicit phonetic instruction. In the second activity (mirror activity), students

were given a mirror to imitate articulatory gestures such as the lip rounding and mouth

opening as shown on the photos in the first activity. The instructor asked them to practice

pronouncing the articles un, une, le, and la by themselves while looking into the mirror.

The segmental-level perception and production phase included three activities.

The first two activities were similar to those in Lee and Lyster (2016a). In the first

activity (pick-up-a-card activity), the students were given two pieces of paper. One had

un on one side and une on the other side. On the other paper, le appeared on one side and

la on the other side. The instructor said one of the sounds and asked the students to show

her the card corresponding to what she had said. The instructor was asked to try 12 trials

(4 sounds × 3 repetitions) to confirm their perception of the sounds. In the second activity

(bingo activity), the students were given the following 12 words: un, une, brun, brune,
 39

lundi, lune, la, le, ma, me, ta, and te. The instructor asked them to choose nine words out

of the 12 words and then to write them on their 3×3 Bingo card provided in the booklet.

The instructor chose nine words and said each word to the students, who in turn circled it

on their card. When a student had completed two Bingo lines (vertical, horizontal, or

diagonal lines), he or she could yell Bingo. The instructor then went to the student and

asked him or her to pronounce all circled words. If the student did not make any

pronunciation errors, he or she won the game. On the contrary, if there were any

pronunciation errors, the instructor corrected them explicitly (not ‘X, but ‘Y’), and then

she continued the game until somebody said Bingo. There were two rounds of this

activity.

Finally, in the third activity (imitation activity), the instructor played the sound

files, previously recorded from native speakers of French, four times. The students were

then required to record their pronunciation of French articles using their cell phone or

computer while imitating the sound files.

There were two sentence-level production activities. In the first activity (choral

repetition activity), there were 12 sentences in the booklet, including French articles

highlighted in red color (e.g., Le Canada est un pays de multiculturalisme). The

instructor read aloud each sentence while exaggerating the articles to draw the students’

attention to the sounds. Once the reading was done, there was choral repetition as

students together read aloud each sentence. In the second activity (dictation activity),

students in pairs (A and B) participated in a mini dictation exercise. Learner A read 10

sentences including the target sounds to Learner B, who in turn completed relevant

blanks, and vice versa. There were one to three blanks in each sentence. The students
 40

were required to write un, une, le, or la in each blank. After the activity, the pair of

students checked their answers with each other.

In the sentence-level perception activity, the students were first given a total of 15

sentences collected from the texts that they had read. The instructor played an audio clip

and then asked the students to find any pronunciation errors in each sentence. For

instance, one of the 15 sentences in the booklet was written as “Il y a toujours une

réponse à une question”. However, the speaker in the audio clip intentionally

mispronounced one article as follows: “Il y a toujours une réponse à un question”. The

instructor played the audio clip twice and made a brief pause between the sentences.

Once it was completed, the instructor asked the students to give their own answers and

confirmed them.

Finally, the students were asked to answer questions in a storytelling format

(spontaneous-level practice), all of which were related to the content of the texts that they

had read. The questions required them to use French articles in their storytelling.

In contrast to Condition 1, concerning unclear pronunciation of French articles,

the instructor explicitly corrected the pronunciation by providing a model of correct

pronunciation (e.g., not ‘X, but ‘Y’) and an opportunity to repair it.

3. 4. 4. Condition 3: Control Condition (Classes 5 and 6)

The students in the control condition engaged in the regular instruction excluding

any FFI components adopted in the first and second conditions. As part of the course

curriculum, the instructors in this condition provided their students with a list of noun

endings along with their predictability in gender attribution. However, the instructors
 41

offered minimal exercises (only 10 nouns) after presenting the list of the noun endings to

their students only once.

3. 5. Measures

3. 5. 1. Overview

In order to investigate the effects of the two FFI conditions on the acquisition of

French grammatical gender, a pretest, an immediate posttest, and a delayed posttest were

administered. A total of six tasks were included in each test. First, grammatical judgment

and text-completion tasks measured the participants’ knowledge of French gender

attribution. Given that these tasks induced students to retrieve grammatical knowledge

(i.e., whether a given noun is masculine or feminine) without verbally producing French

articles, the two tasks were adopted to measure grammatical accuracy. Second, forced-

choice identification and read-aloud tasks assessed the extent to which the participants

perceived and produced the sounds of the French articles un, une, le, and la. The forced-

choice identification task measured the degree to which they perceptually categorized the

four sounds. The participants were also asked to verbalize the four sounds twice in the

read-aloud task. In contrast to the grammatical judgment and text-completion tasks, these

two tasks mainly required them to retrieve their phonological knowledge regarding the

sounds without having to account for the noun endings targeted by the FFI on sublexical

cues.

In the picture-description task, the participants were asked to describe pictures in

French, using target nouns and distractors in the singular form. Therefore, they were

required to determine the grammatical gender of the nouns and then verbally produce the
 42

articles followed by the nouns to complete the task successfully. With respect to the

article-noun congruent/incongruent task, they listened to correct and incorrect noun

phrases (e.g., un chapeau and *une chapeau) via a headset and then judged whether they

were grammatically correct or not. To complete this task in a targetlike manner, the

students needed to know the grammatical gender of given nouns in addition to being able

to accurately perceive the sounds of the articles.

There were two versions of each task, A and B, allowing for a counterbalanced

design to reduce practice effects. The participants were randomly selected so that half

followed an ABA sequence across the three testing sessions, while the other half

followed a BAB sequence. For the baseline participants, half of them were given A and

the other half of them received B.

The following was the order of the six tasks in each testing session: (1)

grammatical judgment task, (2) text-completion task, (3) picture-description task, (4)

read-aloud task, (5) forced-choice identification task, and (6) article-noun

congruent/incongruent task.

The students also completed three tests measuring their EF skills to examine the

extent to which learning gains were mediated by their cognitive controls. A different

measure was administered at each testing time: the Simon Test with the pretest; the Corsi

Block-Tapping Test with the immediate posttest; the Wisconsin Card Sorting Test with

the delayed posttest. The rationale for administering them at different times was to ensure

that the students completed each test within given class hours. The Simon Test was

designed to measure their inhibitory control. The Corsi Block-Tapping Test was
 43

administered to measure their nonverbal visuospatial working memory. The Wisconsin

Card Sorting Test was conducted to measure their cognitive flexibility.

3. 5. 2. Target Nouns

Based on the target noun endings presented in Table 3, various inanimate nouns

were used in the grammatical judgment, text-completion, picture-description, and article-

noun congruent/incongruent tasks (refer to each task for details). Given that the

pronunciation of un changes before a vowel, all nouns began with consonants, which also

controlled for the use of the gender-neutral elided article l’.

The target nouns were categorized as familiar nouns, high-frequency unfamiliar

nouns, and low-frequency unfamiliar nouns. The familiar nouns appeared in the

instructional materials, whereas the unfamiliar nouns did not. In particular, the high-

frequency unfamiliar nouns were the ones used frequently in many French L2 textbooks

and on a daily basis. Given the L2 participants’ proficiency level, therefore, the

participants were likely to have encountered them before, but the nouns did not occur

during the instructional sessions and were not explicitly taught. They were unlikely to

know and learn the low-frequency unfamiliar nouns due to their low frequency.

According to Lexique 3.82 (New, Pallier, Ferrand, & Matos, 2001), the high-

frequency unfamiliar nouns in the tests were, overall, frequent to very frequent

(frequency M = 39.4; SD = 5.64) in its corpus (5 = very rare, 10 = rare, 20 = frequent, 50

= very frequent). On the other hand, the low-frequency unfamiliar nouns were rare to

very rare (frequency M = 8.23; SD = 9.12). Considering the nature of the tasks, the low-

frequency unfamiliar nouns were only used for the grammatical judgment task.
 44

3. 5. 3. Grammatical Judgment Task

In each of the 48 trials comprising the grammatical judgment task, a French noun

appeared on the participant’s computer screen along with two articles (i.e., un vs. une or

le vs. la; half with un vs. une and half with le vs. la), after which the participant was

asked to select the correct article. There was no predetermined time interval between

trials, so participants clicked the next button to move onto next trials. Of the 48 trials with

48 different nouns, 24 were masculine nouns and 24 were feminine nouns. Each set of 24

trials consisted of 12 familiar nouns, six high-frequency unfamiliar nouns, and six low-

frequency unfamiliar nouns. The familiar nouns were prepared by choosing two nouns

from each of the six noun-ending types in Table 3. In a similar manner, one noun from

each of the six noun-ending types was selected for the six high-frequency unfamiliar

nouns and the six low-frequency unfamiliar nouns.

3. 5. 4. Text-Completion Task

For the text-completion task, participants were required to write two separate texts

(i.e., writing an email and a short advertisement), in which they were required to use a

total of 12 nouns (i.e., six masculine and six feminine nouns). The six masculine nouns

included four familiar nouns and two high-frequency unfamiliar nouns.

To control for the number of trials per task, not all noun endings could be

included, so they were selected on the basis of their frequency in the instructional

materials (see Table 3). The four familiar nouns included nouns associated with noun-

ending Types 1 to 4 in Table 3: one from ‘-ant’ or ‘-ent’; one from ‘-eau’, ‘-ot’, or ‘-o’;

one from ‘-ais’, ‘-ait’, ‘-et’, or ‘-ès’; one from ‘-our’, ‘-oir’, or ‘-ort’. The two high-

frequency unfamiliar nouns included nouns associated with noun-ending Types 1 to 2:

 45

one from ‘-ant’ or ‘-ent’; one from ‘-eau’, ‘-ot’, or ‘-o’. The same method was applied for

the six feminine nouns.

Given that this task was not intended to measure whether the participants knew

the meanings of the nouns, the English equivalent of each French noun was provided

(e.g., billet – ticket). The participants were given 10 minutes (i.e., a timed setting) and

two verbs per task were also provided. Finally, it is also important to state that the

participants were explicitly instructed to use all nouns in the singular form.

3. 5. 5. Forced-Choice Identification Task

Audio stimuli were first prepared with the aid of the two audio-stimuli speakers.

The speakers were asked to utter the four French articles (i.e., un, une, le, and la), which

were audio-recorded in the research office. During the test, the participants listened to a

stimulus and then were asked to select what they heard among four options (i.e., un, une,

le, and la). Each stimulus was played only once; there was no predetermined time interval

between trials and the participants moved on to the next trials by clicking the next button

on the computer screen. They completed a total of 32 trials (4 words × 2 speakers × 4

repetitions).

3. 5. 6. Read-Aloud Task

Participants were asked to pronounce the four sounds (i.e., un, une, le, and la)

twice (Attempt 1 and Attempt 2). Their productions were audio-recorded via a headset.

3. 5. 7. Picture-Description Task

Participants were instructed to describe six pictures, each of which included four

target nouns (i.e., 24 nouns including 12 masculine and 12 feminine nouns). The 12

masculine nouns comprised eight familiar nouns and four high-frequency unfamiliar
 46

nouns. Of the eight familiar nouns, six were selected from each of the six noun-ending

types in Table 3. Two nouns associated with noun-ending Types 1 to 2 (one ‘-ant’ or

‘-ent’; one from ‘-eau’, ‘-ot’, or ‘-o’) were then added based on their high frequency in

the instructional materials. To control for the number of trials per task, not all noun

endings could be included for trials with high-frequency unfamiliar nouns. Based on the

frequency in the instructional materials, therefore, the four high-frequency unfamiliar

nouns included nouns associated with noun-ending Types 1 to 4: one from ‘-ant’ or

‘-ent’; one from ‘-eau’, ‘-ot’, or ‘-o’; one from ‘-ais’, ‘-ait’, ‘-et’, or ‘-ès’; one from

‘-our’, ‘-oir’, or ‘-ort’. The same method was applied for the 12 feminine nouns.

In each picture, there were four target nouns (e.g., un rasoir, un mirroir, une

lotion, and une photographie in Figure 2) in addition to one distractor (e.g., une brosse à

dents). In order to draw the participants’ attention to the target nouns, all singular objects

illustrating the target nouns were identified with a check mark. The participants were

required to read a sheet showing all target nouns including distractors before the task and

then asked to use them to describe the pictures with correct articles (un, une, le, or la).

Their productions were audio-recorded via a headset.

Figure 2. Example of a picture-description task

 47

3. 5. 8. Article-Noun Congruent/Incongruent Task

Participants listened to a noun phrase with its correct or incorrect article (e.g., un

chapeau and *une chapeau) via a headset and were asked to determine whether or not the

audio stimulus was grammatical in French by clicking either the right or wrong button on

the computer screen. Each stimulus was played only once. There was no predetermined

time interval between trials. Therefore, the participants moved on to the next trials by

clicking the next button on the computer screen.

Audio stimuli were first prepared with the two audio-stimuli speakers. The

speakers were asked to produce 48 nouns phrases (i.e., 48 trials), half of which were

recorded by the male speaker and half of which were recorded by the female speaker. The

48 trials were composed of 24 trials with masculine nouns and 24 trials with feminine

nouns. The 24 trials with masculine nouns were prepared with eight familiar nouns and

four high-frequency unfamiliar nouns. The eight familiar nouns were arranged by

selecting one noun from each of the six noun-ending types in addition to two nouns

associated with noun-ending Types 1 to 2 (one ‘-ant’ or ‘-ent’; one from ‘-eau’, ‘-ot’, or

‘-o’) entailing high frequency in the instructional materials. To control for the number of

trials per task, not all noun endings could be included for trials with high-frequency

unfamiliar nouns. The four high-frequency unfamiliar nouns included nouns associated

with noun-ending Types 1 to 4: one from ‘-ant’ or ‘-ent’; one from ‘-eau’, ‘-ot’, or ‘-o’;

one from ‘-ais’, ‘-ait’, ‘-et’, or ‘-ès’; one from ‘-our’, ‘-oir’, or ‘-ort’. Four of the eight

familiar nouns were prepared with un-une; for instance, un stylo (congruent trial) and

*une stylo (incongruent trial), whereas the remaining four familiar nouns were prepared

with le-la such as *le vie (incongruent) and la vie (congruent). In the same vein, two
 48

unfamiliar nouns from the four high-frequency unfamiliar nouns were prepared for un-

une trials, whereas the remaining two unfamiliar nouns were prepared for le-la trials.

Accordingly, there were a total of 24 trials for the masculine nouns (12 congruent and 12

incongruent trials). The same method was applied for the feminine nouns (i.e., 24 trials

including 12 congruent and 12 incongruent trials).

3. 5. 9. Measures of Executive Function Skills

The Simon Test, the Corsi Block-Tapping Test, and the Wisconsin Card Sorting

Test were adopted to measure students’ EF skills. In the Simon Test, the students were

asked to press the a key on the left side of the keyboard in response to the word left on

the computer screen and the l key on the right side of the keyboard in response to the

word right on the computer screen. The words appeared either on the left side of the

screen or on the right side of the screen, which resulted in congruent trials (i.e., the word

left appearing on the left side of the screen; the word right appearing on the right side of

the screen) and incongruent trials (i.e., the word left appearing on the right side of the

screen; the word right appearing on the left side of the screen). Each stimulus was present

on the screen until the participant responded up to a maximum of 5,000 ms. Participants

were required to complete practice trials until they answered eight consecutive trials

correctly, after which a total of 28 trials (14 congruent and 14 incongruent trials) were

provided in a randomized order. The Simon Test measured response times on congruent

and incongruent trials.

In the Corsi Block-Tapping Test, stimuli were a random array of blocks spread

out on the computer screen. In each trial, each block flashed one at a time in a sequence.

The students were required to repeat the sequence in the same order by clicking each
 49

block using a computer mouse. The sequences began with two blocks. There were two

trials for each sequence length, and the sequences increased by one block after every

second trial. Testing was terminated when the students failed to complete both trials in a

given sequence. The students had a practice session beforehand. The test recorded the

longest sequence length that a student could replicate.

In the Wisconsin Card Sorting Test, the students were asked to match a given card

to one of the four cards presented on the computer screen. They were required to find a

matching rule (i.e., by color, shape, or number of shapes) by themselves while

completing a few trials with right-or-wrong feedback. Once they correctly matched a set

number of consecutive cards using one matching rule, a new matching rule was

introduced to the students who were in turn requested to find the new rule and complete

another set of trials. Testing was terminated once they correctly matched cards in six

categories or 128 cards. The test recorded preservation errors which resulted from

upholding the previous matching rule in spite of its change.

3. 6. Data Preparation

Concerning the grammatical judgment, forced-choice identification, and article-

noun congruent/incongruent tasks, each score was prepared by calculating percentages of

correct responses. Owing to the small number of items (eight familiar and four high-

frequency unfamiliar nouns), the number of correct responses was counted for the text-

completion task. For the grammatical judgment task, each score per participant was

calculated by lexical familiarity and frequency (i.e., familiar, high-frequency unfamiliar,

and low-frequency unfamiliar nouns) on each test. With respect to the text-completion
 50

and article-noun congruent/incongruent tasks, each score per participant was prepared by

lexical familiarity (i.e., familiar and high-frequency unfamiliar nouns) on each test. As

for the forced-choice identification task, each score per participant was calculated on

each test.

Regarding the read-aloud and picture-description tasks, the target sounds audio-

recorded by the participants were first extracted from raw speech files, each of which was

rated by the six NS raters. For instance, the raters were asked to judge whether each

sound file referred to un, une, le, la, or aucun/pas sûr after listening to the file collected

from the read-aloud task, and then to assess how good the pronunciation was between 1

(Strongly/extremely difficult to understand) and 9 (Strongly/extremely easy to

understand). For the sounds from the picture-description task, the raters were instructed

to choose the article they believed the participant used before the target noun (e.g., un,

une, le, la, or aucun/pas sûr) and then to score it between 1 (Strongly/extremely difficult

to understand) and 9 (Strongly/extremely easy to understand). The raters were allowed to

listen to a stimulus as often as needed before moving on to the next stimulus by clicking

the next button on the computer screen. If the raters chose the correct response (e.g.,

choosing une when a student was instructed to pronounce une in the read-aloud task), its

rating (1 to 9) was taken into account for further analyses. Otherwise (e.g., incorrect

responses or aucun/pas sûr), ‘0’ was recorded regardless of its rating (see also Lee &

Lyster, 2017).

Given that one sound file was rated by the six NS raters, interrater agreement

(Cronbach’s alpha) was calculated. All speech samples indicated reliability indexes of

0.70-0.80, which are considered acceptable in L2 research (Larson-Hall, 2010).

 51

Therefore, by averaging the six NS raters, each score per participant was calculated by

article (i.e., un, une, le, and la) for the read-aloud task at each testing session. Similarly,

for the picture-description task, each score per participant was prepared by lexical

familiarity (familiar and high-frequency unfamiliar nouns) at each testing session.

There were missing data exclusively from the text-completion and picture-

description tasks in cases where the participants did not use a sufficient number of target

nouns in their writing and oral production. The missing data accounted for less than 5%

in the entire data set and were excluded for subsequent analyses.

Regarding the Simon Test, a Simon effect score was calculated for each

participant by subtracting mean response times on congruent trials from mean response

times on incongruent trials. As shown in previous studies (e.g., Bialystok, Craik, Klein, &

Viswanathan, 2004), a smaller Simon effect score refers to higher inhibitory control.

With respect to the Corsi Block-Tapping Test, the score was the longest sequence length

that a participant could replicate (Milner, 1971). Finally, of particular interest in the

Wisconsin Card Sorting Task is the percentage of preservation errors resulting from

upholding the previous matching rule in spite of its change. The percentage of

preservation errors is known to be an indicator of cognitive shifting ability (Miyake et al.,

2000).

3. 7. Summary

This chapter introduced the participants, the design and procedures of the study,

and data preparation. Instructional treatments were described, after which six linguistic

measures and three EF measures were presented. In addition, the chapter also explained
 52

the procedure of data preparation for data analysis. Next, Chapter 4 reports the data

analysis and results.

 53

Chapter 4

Results

This chapter presents the results of the current study. After presenting the data

collected from the baseline participants and the students in the control condition, the

chapter entails three main sections: (a) the effects of two FFI conditions, (b) the variables

affecting the performance of the picture-description task, and (c) the roles of EF skills in

learning gains. Each section begins with its statistical model followed by results. The

chapter concludes by summarizing the results of the present study.

4. 1. Pre-Analysis

As expected, the L1 baseline participants showed maximum scores on all

measures. In the grammatical judgment task, they attained mean accuracy scores of 98.33

(SD = .82), 99.86 (SD = .76), and 99.61 (SD = .91) for the familiar, high-frequency

unfamiliar, and low-frequency unfamiliar nouns (out of 100), respectively. They also

attained mean accuracy scores of 7.97 (SD = .18) and of 3.97 (SD = .18) for the familiar

(out of 8) and high-frequency unfamiliar nouns (out of 4) in the text-completion task.

In the forced-choice identification task, their mean accuracy score was 98.33 (SD

= 1.78) out of 100. Their mean ratings in the read-aloud task were 8.90 (SD = .40) for un,

8.97 (SD = .18) for une, 8.93 (SD = .25) for le, and 8.97 (SD = .18) for la (out of 9).

Moreover, the baseline participants attained high pronunciation ratings in the picture-

description task (M = 8.90 out of 9, SD = .31 for the familiar nouns; M = 8.90 out of 9,

SD = .32 for the high-frequency unfamiliar nouns). Their mean accuracy score in the
 54

article-noun congruent/incongruent task was 98.12 (SD = 2.79) for familiar nouns and

98.24 (SD = 1.29) for high-frequency unfamiliar nouns (out of 100).

Given the homogeneity of these high results, therefore, it was confirmed that the

tasks required the baseline participants to use their L1 knowledge of gender attribution

and the sounds of French articles rather than nonlinguistic problem-solving skills, which

would have yielded more variable results. In addition, considering that their speech

samples received high ratings from the NS raters, it was also confirmed that the rating

scores from the NS raters were reliable enough to answer the research questions in the

current study.

Before investigating the effects of the two FFI conditions, separate mixed effects

models with repeated measures were conducted to examine whether those in the control

condition showed any significant improvement across the three testing sessions. The

students in the control condition made no significant improvement on any measures (pS >

.05). Their descriptive statistics are reported in the section below along with the two FFI

conditions.

4. 2. Effects of Two Form-Focused Instructional Conditions

4. 2. 1. Statistical Model

The L2 participants’ scores were statistically analyzed using linear mixed effects

models in R (R Core Team, 2016) using the lme4 package (version 1.1-15) and restricted

maximum likelihood.

For the grammatical judgment task, fixed effects included ‘condition’ (FFI on

only sublexical cues, FFI on both sublexical cues and pronunciation, and no FFI), ‘time’
 55

(pretest, immediate posttest, and delayed posttest), ‘trial’ (familiar, high-frequency

unfamiliar, and low-frequency unfamiliar nouns), and their two-way and three-way

interactions. The fixed effect factors ‘condition’ and ‘time’ were coded using treatment

coding while having the control condition (condition) and the pretest (time) as a reference

level. The fixed effect factor ‘trial’ was coded using Helmert coding. Therefore, there

were two contrasts: (a) familiar nouns vs. unfamiliar (high-frequency and low-frequency)

nouns; and (b) high-frequency unfamiliar nouns vs. low-frequency unfamiliar nouns. In a

similar vein, for the picture-description and article-noun congruent/incongruent tasks, the

models included the fixed effect factors ‘condition’, ‘time’, ‘trial’ (familiar and high-

frequency unfamiliar nouns), and their two-way and three-way interactions. Considering

that the fixed effect factor ‘trial’ had two categories, the factor was thus centered (-0.5

and 0.5) using the rescale() function in the arm package in R (R Core Team, 2016).

To analyze data for the read-aloud task, two models were designed, each of which

had the fixed effect factors ‘condition’, ‘time’, ‘trial’, and their two-way and three-way

interactions. The first model had two categories (un and une) as its ‘trial’ factor, whereas

the second model had two categories (le and la) as its ‘trial’ factor. For both models, the

fixed effect factor ‘trial’ was centered (-0.5 and 0.5) using the rescale() function in the

arm package in R (R Core Team, 2016). Due to the small number of items in the text-

completion task (eight familiar nouns and four high-frequency unfamiliar nouns), the

number of correct responses—instead of calculating percentages—was counted.

Therefore, there were two models (one for familiar nouns and the other for high-

frequency unfamiliar nouns), each of which had ‘condition’, ‘time’, and their two-way
 56

interactions as fixed effect factors. For the forced-choice identification task, the model

included ‘condition’, ‘time’, and their two-way interactions as fixed effect factors.

For all models, students and classes were treated as random effects with students

nested under classes. Random intercepts for students and classes were included, as were

random slopes for ‘time’ for both students and classes, using a maximal random effects

structure. In addition, with respect to the models including the ‘trial’ factor, random

slopes for ‘trial’ and ‘time’ by ‘trial’ interaction were also included for both students and

classes.

Considering that two-way or three-way interactions including both ‘condition’

and ‘time’ factors were the main interests in the current study, only the interactions

including both factors are interpreted in this chapter. Prior to each analysis, statistical

assumptions were verified (e.g., the explanatory variables were linearly related to the

response; the errors had constant variance, which were independent and normally

distributed). All statistical outcomes were interpreted with alpha set at .05. The intercept

in each model refers to the expected mean value of a dependent variable (y) when all

independent variables (x) are equal to 0.

4. 2. 2. Grammatical Judgment Task

Table 4 summarizes the descriptive statistics for the grammatical judgment task

(see Figure 3 for its boxplot). The maximum score is 100 in the grammatical judgment

task as a result of calculating the percentages of correct responses.

 57

Table 4

Mean Percentage Scores and Standard Deviations (in parentheses) for the Grammatical

Judgment Task

Familiar High-frequency Low-frequency

Condition Test
nouns unfamiliar nouns unfamiliar nouns
66.60 68.75 66.50
Pretest
FFI on only (20.12) (19.39) (22.62)
sublexical cues Immediate 92.46 92.82 93.24
(n = 41; posttest (4.45) (5.10) (5.25)
Condition 1) Delayed 92.19 92.72 92.25
posttest (4.43) (5.49) (6.44)
68.04 68.04 69.17
FFI on both Pretest
(17.86) (18.40) (19.13)
sublexical cues
Immediate 92.21 93.44 92.76
and pronunciation
posttest (4.49) (5.25) (6.25)
(n = 49;
Delayed 93.02 93.60 92.33
Condition 2)
posttest (5.00) (5.51) (6.33)
64.51 66.95 65.60
Pretest
(20.13) (18.32) (16.21)
Control
Immediate 64.77 65.53 65.45
(n = 50;
posttest (16.25) (14.88) (14.68)
Condition 3)
Delayed 67.30 64.69 65.65
posttest (15.11) (16.92) (17.05)

The statistical model summarized in Table 5 shows that the scores of students in

both FFI conditions became significantly higher on the immediate and delayed posttests

compared to the reference levels (i.e., control condition and pretest). Given that there

were no significant three-way interactions that included ‘trial’, it was confirmed that the

students in both FFI conditions significantly improved their scores in the grammatical

judgment task regardless of lexical familiarity and frequency (see also the descriptive

statistics in Table 4).

 58

Table 5

Summary of Fixed Effects for the Statistical Model for the Grammatical Judgment Task

Estimate Standard
Predictor t-value p-value
(β) error
Intercept 65.64 4.19 15.65 .001
Condition 1 2.30 6.04 0.38 .730
Condition 2 2.89 5.96 0.49 .663
Time 2 -0.38 4.39 -0.09 .937
Time 3 0.25 4.33 0.06 .959
Trial 1 -1.18 1.10 -1.07 .286
Trial 2 -0.95 1.32 -0.72 .471
Condition 1×Time 2 25.10 6.31 3.98 .032
Condition 2×Time 2 24.65 6.24 3.95 .035
Condition 1×Time 3 24.08 6.25 3.85 .033
Condition 2×Time 3 24.10 6.16 3.91 .035
Condition 1×Trial 1 0.50 1.65 0.31 .761
Condition 2×Trial 1 0.79 1.59 0.50 .620
Condition 1×Trial 2 -0.63 1.98 -0.32 .752
Condition 2×Trial 2 1.74 1.91 0.91 .361
Time 2×Trial 1 0.69 1.51 0.46 .647
Time 3×Trial 1 2.60 1.51 1.72 .086
Time 2×Trial 2 0.90 1.85 0.48 .629
Time 3×Trial 2 1.63 1.85 0.88 .380
Condition 1×Time 2×Trial 1 -0.38 2.28 -0.17 .869
Condition 2×Time 2×Trial 1 -0.89 2.18 -0.41 .684
Condition 1×Time 3×Trial 1 -2.09 2.29 -0.91 .362
Condition 2×Time 3×Trial 1 -2.17 2.18 -1.00 .319
Condition 1×Time 2×Trial 2 0.98 2.79 0.35 .725
Condition 2×Time 2×Trial 2 -2.17 2.66 -0.82 .415
Condition 1×Time 3×Trial 2 -0.37 2.80 -0.13 .895
Condition 2×Time 3×Trial 2 -3.29 2.66 -1.24 .217
Note. p-values smaller than .05 are highlighted in bold. Trial 1 is a comparison between
familiar nouns and unfamiliar (high-frequency and low-frequency) nouns, whereas Trial
2 is a comparison between high-frequency unfamiliar nouns and low-frequency
unfamiliar nouns.

4. 2. 3. Text-Completion Task

Table 6 includes descriptive statistics for the text-completion task (see Figure 3

for its boxplot). The maximum score is 8 for the familiar nouns and 4 for the high-

frequency unfamiliar nouns as a result of counting the raw number of correct responses.
 59

Table 6

Mean Scores and Standard Deviations (in parentheses) for the Text-Completion Task

Familiar High-frequency
Condition Test
nouns unfamiliar nouns
2.43 1.10
Pretest
FFI on only (1.90) (1.18)
sublexical cues Immediate 4.39 1.67
(n = 41; posttest (1.80) (1.40)
Condition 1) Delayed 4.95 1.70
posttest (1.84) (1.32)
2.41 1.15
FFI on both Pretest
(1.42) (1.03)
sublexical cues
Immediate 4.31 1.58
and pronunciation
posttest (1.74) (1.29)
(n = 49;
Delayed 4.71 1.81
Condition 2)
posttest (2.03) (1.44)
2.16 1.12
Pretest
(1.77) (1.12)
Control
Immediate 2.84 1.22
(n = 50;
posttest (2.10) (1.22)
Condition 3)
Delayed 2.30 1.22
posttest (2.10) (1.09)

According to Table 7 summarizing the statistical outcomes, with respect to the

familiar nouns, the scores of students in both FFI conditions were significantly higher on

the immediate and delayed posttests compared to the reference levels. However, there

were no significant interactions for the high-frequency unfamiliar nouns in the text-

completion task.
 60

Table 7

Summary of Fixed Effects for the Statistical Model for the Text-Completion Task

Estimate Standard
Trial Predictor t-value p-value
(β) error
Intercept 2.23 0.26 8.30 < .001
Condition 1 0.26 0.39 0.67 .503
Condition 2 0.22 0.37 0.59 .558
Time 2 0.68 0.30 2.28 .023
Familiar
Time 3 0.80 0.30 2.68 .008
nouns
Condition 1×Time 2 1.32 0.45 2.93 .004
Condition 2×Time 2 1.23 0.43 2.85 .005
Condition 1×Time 3 1.72 0.45 3.81 < .001
Condition 2×Time 3 1.53 0.43 3.56 < .001
Intercept 1.20 0.18 6.13 < .001
Condition 1 -0.08 0.27 -0.29 .776
Condition 2 0.03 0.26 0.11 .912
High-
Time 2 0.10 0.21 0.46 .640
frequency
Time 3 0.10 0.21 0.46 .640
unfamiliar
Condition 1×Time 2 0.54 0.32 1.67 .095
nouns
Condition 2×Time 2 0.33 0.31 1.07 .281
Condition 1×Time 3 0.51 0.32 1.57 .116
Condition 2×Time 3 0.56 0.31 1.81 .071
Note. p-values smaller than .05 are highlighted in bold.

Grammatical Judgment Task Text-Completion Task

Figure 3. Boxplots of the scores of grammatical measures (Condition 1 = FFI on only

sublexical cues; Condition 2 = FFI on both sublexical cues and pronunciation; Condition
3 = Control; Trial 1 = Familiar nouns; Trial 2 = High-frequency unfamiliar nouns; Trial 3
= Low-frequency unfamiliar nouns).
 61

4. 2. 4. Forced-Choice Identification Task

The descriptive statistics for the forced-choice identification task are reported in

Table 8 (see Figure 4 for its boxplot). The maximum score is 100 in the forced-choice

identification task as a result of calculating the percentages of correct responses.

Table 8

Mean Scores and Standard Deviations (in parentheses) for the Forced-Choice

Identification Task

Condition Test Score

98.11
Pretest
FFI on only (8.23)
sublexical cues Immediate 99.30
(n = 41; posttest (1.62)
Condition 1) Delayed 99.44
posttest (1.84)
98.35
FFI on both Pretest
(5.41)
sublexical cues
Immediate 99.43
and pronunciation
posttest (1.35)
(n = 49;
Delayed 99.12
Condition 2)
posttest (2.57)
96.00
Pretest
(15.4)
Control
Immediate 98.91
(n = 50;
posttest (3.56)
Condition 3)
Delayed 96.85
posttest (14.28)

The students in all conditions showed maximum scores on all tests. As shown in

Table 9, the statistical model did not detect any significant interactions in the forced-

choice identification task.

 62

Table 9

Summary of Fixed Effects for the Statistical Model for the Forced-Choice Identification

Task

Estimate Standard
Predictor t-value p-value
(β) error
Intercept 97.15 1.44 66.41 < .001
Condition 1 2.12 2.12 1.00 .342
Condition 2 2.45 2.07 1.18 .268
Time 2 2.91 1.63 1.79 .075
Time 3 0.85 1.63 0.52 .602
Condition 1×Time 2 -1.73 2.45 -0.71 .481
Condition 2×Time 2 -1.83 2.34 -0.78 .434
Condition 1×Time 3 0.48 2.46 0.20 .844
Condition 2×Time 3 -0.09 2.34 -0.04 .971
Note. A p-value smaller than .05 is highlighted in bold.

4. 2. 5. Read-Aloud Task

The descriptive statistics for the read-aloud task appear in Table 10 (see Figure 4

for its boxplot). The maximum score is 9 for each sound in the read-aloud task, indicating

1 (Strongly/extremely difficult to understand) to 9 (Strongly/extremely easy to

understand).
 63

Table 10

Mean Scores and Standard Deviations (in parentheses) for the Read-Aloud Task

Condition Test un une le la

4.56 5.95 5.98 7.46
Pretest
FFI on only (2.34) (1.98) (2.04) (1.35)
sublexical cues Immediate 4.20 5.87 7.42 7.74
(n = 41; posttest (1.39) (1.46) (1.33) (1.18)
Condition 1) Delayed 4.10 5.92 7.15 7.59
posttest (1.38) (1.09) (1.40) (1.15)
4.36 5.22 6.22 6.45
FFI on both Pretest
(1.87) (2.34) (1.91) (2.10)
sublexical cues
Immediate 7.84 7.73 7.38 7.55
and pronunciation
posttest (1.21) (1.24) (.70) (.91)
(n = 49;
Delayed 7.52 7.61 6.60 6.83
Condition 2)
posttest (1.57) (1.37) (2.05) (2.10)
4.49 6.34 6.37 7.43
Pretest
(1.86) (1.59) (1.48) (1.28)
Control
Immediate 4.92 6.57 7.44 7.79
(n = 50;
posttest (1.57) (1.75) (1.00) (.43)
Condition 3)
Delayed 4.66 6.35 7.14 7.51
posttest (1.62) (1.64) (1.16) (.86)

The statistical model in Table 11 reveals that the students receiving FFI on both

sublexical cues and pronunciation showed significant improvement on the immediate and

delayed posttests compared to the reference levels. Moreover, considering that the three-

way interactions with ‘trial’ failed to reach significance, there were no significant

differences between un and une in terms of the improvement. In contrast, those receiving

FFI on only sublexical cues did not show any significant two-way and three-way

interactions.
 64

Table 11

Summary of Fixed Effects for the Statistical Model for un and une

Estimate Standard
Predictor t-value p-value
(β) error
Intercept 5.40 0.70 7.73 .005
Condition 1 -0.11 0.99 -0.11 .920
Condition 2 -0.72 0.99 -0.73 .519
Time 2 0.36 0.52 0.69 .541
Time 3 0.12 0.55 0.22 .839
Trial 1.91 0.51 3.72 .023
Condition 1×Time 2 -0.62 0.74 -0.83 .468
Condition 2×Time 2 2.70 0.75 3.61 .035
Condition 1×Time 3 -0.39 0.77 -0.50 .650
Condition 2×Time 3 2.71 0.78 3.46 .039
Condition 1×Trial -0.55 0.73 -0.76 .492
Condition 2×Trial -1.05 0.74 -1.42 .229
Time 2×Trial -0.21 0.29 -0.74 .461
Time 3×Trial -0.16 0.29 -0.55 .584
Condition 1×Time 2×Trial 0.51 0.41 1.24 .218
Condition 2×Time 2×Trial -0.76 0.43 -1.76 .079
Condition 1×Time 3×Trial 0.62 0.41 1.50 .135
Condition 2×Time 3×Trial -0.60 0.43 -1.38 .170
Note. p-values smaller than .05 are highlighted in bold.

Table 12 summarizes fixed factors for the statistical model for the sounds le and

la. The analysis indicated significant three-way interactions on both posttests in the group

receiving FFI on both sublexical cues and pronunciation. In other words, their scores

were significantly higher on the immediate and delayed posttests compared to the

reference levels, particularly favoring the sound la. However, the analysis revealed no

significant interactions in the group receiving FFI on only sublexical cues.

 65

Table 12

Summary of Fixed Effects for the Statistical Model for le and la

Estimate Standard
Predictor t-value p-value
(β) error
Intercept 6.88 0.35 19.64 < .001
Condition 1 -0.14 0.50 -0.28 .795
Condition 2 -0.58 0.51 -1.14 .329
Time 2 0.74 0.42 1.78 .178
Time 3 0.45 0.50 0.90 .438
Trial 1.06 0.18 6.03 .001
Condition 1×Time 2 0.10 0.59 0.18 .873
Condition 2×Time 2 0.43 0.60 0.72 .522
Condition 1×Time 3 0.18 0.71 0.25 .818
Condition 2×Time 3 -0.02 0.72 -0.03 .981
Condition 1×Trial 0.40 0.25 1.59 .156
Condition 2×Trial -0.85 0.26 -3.28 .012
Time 2×Trial -0.71 0.18 -3.86 .000
Time 3×Trial -0.69 0.18 -3.80 .000
Condition 1×Time 2×Trial -0.45 0.26 -1.71 .089
Condition 2×Time 2×Trial 0.65 0.28 2.37 .019
Condition 1×Time 3×Trial -0.34 0.26 -1.28 .203
Condition 2×Time 3×Trial 0.71 0.28 2.57 .011
Note. p-values smaller than .05 are highlighted in bold.

Forced-Choice Identification Task Read-aloud Task

Figure 4. Boxplots of the scores of phonological measures (Condition 1 = FFI on only

sublexical cues; Condition 2 = FFI on both sublexical cues and pronunciation; Condition
3 = Control).
 66

4. 2. 6. Picture-Description Task

Table 13 includes descriptive statistics for the picture-description task (see Figure

5 for its boxplot). The maximum score is 9 in the picture-description task, indicating 1

(Strongly/extremely difficult to understand) to 9 (Strongly/extremely easy to understand).

Table 13

Mean Scores and Standard Deviations (in parentheses) for the Picture-Description Task

Familiar High-frequency
Condition Test
nouns unfamiliar nouns
1.63 1.42
Pretest
FFI on only (1.00) (.97)
sublexical cues Immediate 2.61 2.43
(n = 41; posttest (1.22) (1.33)
Condition 1) Delayed 2.84 2.94
posttest (1.27) (1.58)
1.62 1.34
FFI on both Pretest
(1.03) (1.13)
sublexical cues
Immediate 6.86 2.58
and pronunciation
posttest (1.31) (1.75)
(n = 49;
Delayed 7.07 2.81
Condition 2)
posttest (1.39) (1.65)
1.97 1.59
Pretest
(.83) (1.10)
Control
Immediate 2.33 2.25
(n = 50;
posttest (1.14) (1.32)
Condition 3)
Delayed 2.35 2.46
posttest (.91) (1.25)

As shown in Table 14, the students receiving FFI on both sublexical cues and

pronunciation obtained significantly higher scores on the immediate and delayed posttests

(i.e., significant two-way interactions) compared to the reference levels. In particular,

considering that there were also significant three-way interactions including ‘trial’, they

gained significantly higher scores for the familiar nouns in contrast to the high-frequency
 67

unfamiliar nouns. On the other hand, those receiving FFI on only sublexical cues did not

show any significant interactions.

Table 14

Summary of Fixed Effects for the Statistical Model for the Picture-Description Task

Estimate Standard
Predictor t-value p-value
(β) error
Intercept 1.77 0.18 9.91 .001
Condition 1 -0.26 0.26 -1.00 .372
Condition 2 -0.27 0.25 -1.07 .350
Time 2 0.52 0.19 2.79 .037
Time 3 0.65 0.25 2.63 .073
Trial -0.38 0.16 -2.36 .019
Condition 1×Time 2 0.49 0.28 1.79 .122
Condition 2×Time 2 2.69 0.27 10.08 < .001
Condition 1×Time 3 0.68 0.36 1.89 .139
Condition 2×Time 3 2.81 0.35 8.04 .003
Condition 1×Trial 0.18 0.24 0.72 .470
Condition 2×Trial 0.11 0.23 0.49 .625
Time 2×Trial 0.30 0.21 1.40 .162
Time 3×Trial 0.49 0.21 2.27 .024
Condition 1×Time 2×Trial -0.29 0.32 -0.89 .376
Condition 2×Time 2×Trial -4.32 0.31 -14.03 < .001
Condition 1×Time 3×Trial -0.19 0.32 -0.60 .552
Condition 2×Time 3×Trial -4.49 0.31 -14.57 < .001
Note. p-values smaller than .05 are highlighted in bold.
 68

4. 2. 7. Article-Noun Congruent/Incongruent Task

Table 15 summarizes descriptive statistics for the article-noun

congruent/incongruent task (see Figure 5 for its boxplot). The maximum score is 100 in

the article-noun congruent/incongruent task, as a result of calculating the percentages of

correct responses.

Table 15

Mean Scores and Standard Deviations (in parentheses) for the Article-Noun

Congruent/Incongruent Task

Familiar High-frequency
Condition Test
nouns unfamiliar nouns
64.45 63.13
Pretest
FFI on only (14.65) (18.68)
sublexical cues Immediate 84.52 84.26
(n = 41; posttest (12.77) (13.33)
Condition 1) Delayed 88.03 81.74
posttest (11.29) (15.97)
63.72 64.65
FFI on both Pretest
(12.32) (16.39)
sublexical cues
Immediate 88.28 83.07
and pronunciation
posttest (9.84) (12.16)
(n = 49;
Delayed 86.12 84.71
Condition 2)
posttest (12.25) (13.27)
63.94 62.63
Pretest
(16.38) (18.06)
Control
Immediate 66.13 67.50
(n = 50;
posttest (12.23) (16.85)
Condition 3)
Delayed 66.13 61.63
posttest (16.10) (18.90)

Table 16 shows that the students in both FFI conditions attained significantly

higher scores on the immediate and delayed posttests compared to the reference levels.

Moreover, the scores from those receiving FFI on both sublexical cues and pronunciation

had a significant Condition 2×Time 2×Trial interaction. That is, they attained
 69

significantly higher scores for the familiar nouns in comparison to the high-frequency

unfamiliar nouns on the immediate posttest.

Table 16

Summary of Fixed Effects for the Statistical Model for the Article-Noun

Congruent/Incongruent Task

Estimate Standard
Predictor t-value p-value
(β) error
Intercept 63.28 1.90 33.28 < .001
Condition 1 0.28 2.85 0.10 .921
Condition 2 0.92 2.74 0.33 .739
Time 2 3.53 2.03 1.74 .084
Time 3 0.59 2.17 0.27 .785
Trial -1.29 2.02 -0.64 .528
Condition 1×Time 2 17.34 3.05 5.68 < .001
Condition 2×Time 2 17.72 2.92 6.07 < .001
Condition 1×Time 3 20.48 3.28 6.25 < .001
Condition 2×Time 3 20.55 3.12 6.58 < .001
Condition 1×Trial -0.11 3.03 -0.04 .970
Condition 2×Trial 2.26 2.92 0.78 .443
Time 2×Trial 2.69 2.71 0.99 .322
Time 3×Trial -3.19 2.71 -1.18 .240
Condition 1×Time 2×Trial -1.48 4.08 -0.36 .717
Condition 2×Time 2×Trial -8.95 3.89 -2.30 .022
Condition 1×Time 3×Trial -1.71 4.09 -0.42 .677
Condition 2×Time 3×Trial 0.80 3.89 0.20 .838
Note. p-values smaller than .05 are highlighted in bold.
 70

Picture-Description Task Article-Noun Congruent/Incongruent Task

Figure 5. Boxplots of the scores of grammatical + phonological measures (Condition 1 =

FFI on only sublexical cues; Condition 2 = FFI on both sublexical cues and
pronunciation; Condition 3 = Control; Trial 1 = Familiar nouns; Trial 2 = High-frequency
unfamiliar nouns).

4. 3. Variables Affecting the Performance of the Picture-Description Task

4. 3. 1. Statistical Model

Previous studies (e.g., Harley, 1998; Lyster, 2004) reported L2 learners’

ambiguous pronunciation of French articles in oral production (e.g., picture-description

task). In order to investigate the variables affecting the performance of the picture-

description task, multiple linear regression was modelled using the lm() function in R (R

Core Team, 2016). The dependent variable for each participant was an overall score (i.e.,

regardless of lexical familiarity) collected from the picture-description task on each test.

Predictors were five sets of overall scores from the grammatical judgment, text-

completion, read-aloud, forced-choice identification, and article-noun

congruent/incongruent tasks on each test.

There were several sets of multiple linear regression: (a) all students on the

pretest, (b) those in the two FFI conditions on the immediate posttest, and (c) those in the
 71

two FFI conditions on the delayed posttest. The motivation for including the last two

analyses was to examine any changes after the relevant FFI sessions. Before conducting

the analyses, the assumptions of linearity, homoscedasticity, normality, and independence

were verified. All statistical outcomes were interpreted with alpha set at .05.

4. 3. 2. Results

Table 17 summarizes the statistical outcomes of the multiple regression model on

the pretest. The scores of the grammatical judgment task and those of the read-aloud task

were significant predictors of the scores of the picture-description task. The other

predictors failed to reach statistical significance. In particular, the read-aloud task yielded

the standardized β of .34, whereas the grammatical judgment task showed the

standardized β of .22. That is, the read-aloud task had a higher degree of importance in

the model.

Table 17

Linear Model of Predictors of the Performance of the Picture-Description Task on the

Pretest with 95% Confidence Intervals (in parentheses)

Standard
Predictor B β t-value p-value
error (B)
-.86
Intercept .70 -1.23 .220
(-2.24, .52)
.01
Grammatical judgment task .01 .22 2.68 .008
(.01, .02)
.06
Text-completion task .06 .07 .98 .328
(-.06, .18)
Forced-choice identification .01
.01 .01 .023 .982
task (-.02, .02)
.23
Read-aloud task .05 .34 4.44 < .001
(.13, .33)
Article-noun .01
.01 .03 .31 .756
congruent/incongruent task (-.01, .01)
Note. R2 = .21. p-values smaller than .05 are highlighted in bold.
 72

Tables 18 and 19 show statistical outcomes for the two FFI conditions on the

immediate and delayed posttests. After the FFI sessions, students in the two FFI

conditions demonstrated different patterns. The scores of the read-aloud task were the

only significant predictor of those of the picture-description task on both posttests.

Table 18

Linear Model of Predictors of the Performance of the Picture-Description Task on the

Immediate Posttest with 95% Confidence Intervals (in parentheses)

Standard
Condition Predictor B β t-value p-value
error (B)
8.25
Intercept 8.87 .93 .359
(-9.79, 26.29)
Grammatical .05
.04 .17 1.35 .187
judgment task (-.02, .12)
FFI on only
Text-completion .07
sublexical .05 .17 1.46 .154
task (-.03, .17)
cues
Forced-choice -.14
(n = 41; .09 -.20 -1.61 .117
identification task (-.32, .04)
Condition 1)
.57
Read-aloud task .09 .80 6.39 < .001
(.39, .75)
Article-noun
-.01
congruent/incongruent .01 -.01 -.11 .914
(-.02, .02)
task
-1.50
Intercept 5.27 -.28 .778
(-12.13, 9.15)
Grammatical -.02
.05 -.07 -.36 .719
judgment task (-.13, .09)
FFI on both
Text-completion .05
sublexical .08 .11 .67 .504
task (-.10, .20)
cues and
Forced-choice .02
pronunciation .03 .10 .69 .497
identification task (-.04, .09)
(n = 49;
.47
Condition 2) Read-aloud task .19 .37 2.42 .020
(.08, .85)
Article-noun
.03
congruent/incongruent .02 .21 1.13 .264
(-.02, .08)
task
Note. R2 = .61 for Condition 1. R2 = .18 for Condition 2. p-values smaller than .05 are
highlighted in bold.
 73

Table 19

Linear Model of Predictors of the Performance of the Picture-Description Task on the

Delayed Posttest with 95% Confidence Intervals (in parentheses)

Standard
Condition Predictor B β t-value p-value
error (B)
-4.81
Intercept 12.13 -.40 .695
(-29.51, 19.90)
Grammatical -.01
.07 -.01 -.04 .967
judgment task (-.14, .13)
FFI on only
Text-completion .07
sublexical .08 .15 .87 .388
task (-.10, .24)
cues
Forced-choice .03
(n = 41; .12 .04 .24 .809
identification task (-.22, .28)
Condition 1)
.29
Read-aloud task .10 .44 2.80 .009
(.08, .50)
Article-noun
.03
congruent/incongruent .02 .29 1.23 .228
(-.02, .08)
task
-.63
Intercept 8.27 -.08 .940
(-17.32, 16.10)
Grammatical -.02
.05 -.07 -.37 .713
judgment task (-.13, .09)
FFI on both
Text-completion .10
sublexical .07 .21 1.40 .169
task (-.04, .23)
cues and
Forced-choice .01
pronunciation .08 .02 .17 .865
identification task (-.14, .17)
(n = 49;
.53
Condition 2) Read-aloud task .20 .37 2.63 .012
(.12, .93)
Article-noun
.02
congruent/incongruent .02 .19 .99 .328
(-.02, .07)
task
Note. R2 = .31 for Condition 1. R2 = .24 for Condition 2. p-values smaller than .05 are
highlighted in bold.

4. 4. Roles of Executive Function Skills in Learning Gains

4. 4. 1. Statistical Model

To examine the extent to which the gains made by the L2 participants in each task

were mediated by their EF skills (i.e., inhibitory control, nonverbal visuospatial working
 74

memory, and cognitive flexibility), multiple linear regression was modelled using the

lm() function in R (R Core Team, 2016). There were two models for each task. In the first

model, the dependent variable was immediate gains (i.e., immediate posttest minus

pretest), whereas in the second model the dependent variable was delayed gains (i.e.,

delayed posttest minus pretest). A score for each task at each testing session was prepared

by calculating an overall score from all trials (i.e., regardless of lexical familiarity and

frequency). Given the specific research question, only the students in the two FFI

conditions were included in the analyses. Moreover, the forced-choice identification task

was excluded since all participants showed maximum scores on all tests.

The students’ Simon effect scores (inhibitory control), Corsi spans (nonverbal

visuospatial working memory), and percentages of preservation errors in the Wisconsin

Card Sorting Test (cognitive flexibility) were entered as predictors in the models. Prior to

the analyses, the statistical assumptions such as linearity, homoscedasticity, normality,

and independence were verified. All statistical outcomes were interpreted with alpha set

at .05.

4. 4. 2. Results

The mean of the Simon effect scores was 58.68 (SD = 66.10). The mean of the

Corsi spans was 6.03 (SD = 1.73) and that of the percentages of preservation errors in the

Wisconsin Card Sorting Test was 12.15 (SD = 6.77).

Tables 20 and 21 summarize the linear models of predictors in the grammatical

judgment and text-completion tasks. For the grammatical judgment task, the nonverbal

visuospatial working memory was the only significant predictor of the immediate and

delayed gains. According to the model, the higher Corsi spans (i.e., higher working
 75

memory) the students had, the greater learning gains they showed in the grammatical

judgment task. Yet, none of the EF skills predicted the learning gains for the text-

completion task.

Table 20

Linear Model of Predictors of the Learning Gains of the Grammatical Judgment Task

with 95% Confidence Intervals (in parentheses)

Standard
Gains Predictor B β t-value p-value
error (B)
-21.00
Intercept 7.05 -2.98 .004
(-34.99, -7.01)
Inhibitory .03
.02 .15 1.54 .126
control (-.01, .06)
Immediate Nonverbal
gains visuospatial 2.57
.94 .28 2.74 .007
working (.71, 4.42)
memory
Cognitive .18
.24 .08 .76 .447
flexibility (-.29, .66)
-33.49
Intercept 8.29 -4.04 < .001
(-49.95, -17.04)
Inhibitory .02
.02 .16 1.60 .113
control (-.02, .06)
Delayed Nonverbal
gains visuospatial 4.01
1.10 .36 3.64 < .001
working (1.82, 6.20)
memory
Cognitive .46
.28 .16 1.60 .113
flexibility (-.11, 1.01)
Note. R2 = .10 for the immediate gains. R2 = .13 for the delayed gains. p-values smaller
than .05 are highlighted in bold.
 76

Table 21

Linear Model of Predictors of the Learning Gains of the Text-Completion Task with 95%

Confidence Intervals (in parentheses)

Standard
Gains Predictor B β t-value p-value
error (B)
2.75
Intercept 1.44 1.91 .059
(-.11, 5.61)
Inhibitory .01
.01 .18 1.78 .078
control (-.01, .01)
Immediate Nonverbal
gains visuospatial -.19
.19 -.10 -.97 .334
working (-.57, .19)
memory
Cognitive -.04
.05 -.08 -.74 .460
flexibility (-.13, .06)
1.78
Intercept 1.51 1.18 .240
(-1.21, 4.77)
Inhibitory .01
.01 .09 .86 .393
control (-.01, .01)
Delayed Nonverbal
gains visuospatial -.07
.20 -.04 -.34 .733
working (-.47, .33)
memory
Cognitive .04
.05 .08 .76 .449
flexibility (-.06, .14)
Note. R2 = .05 for the immediate gains. R2 = .02 for the delayed gains.

Table 22 shows the linear model of predictors in the read-aloud task. As shown in

Table 22, the inhibitory control was the only significant predictor of the immediate and

delayed gains. In light of the model, the smaller Simon effect scores (i.e., higher

inhibitory control) the students showed, the greater gains they made in the read-aloud

task.
 77

Table 22

Linear Model of Predictors of the Learning Gains of the Read-Aloud Task with 95%

Confidence Intervals (in parentheses)

Standard
Gains Predictor B β t-value p-value
error (B)
.66
Intercept .70 .94 .347
(-.73, 2.06)
Inhibitory -.01
.01 -.42 -4.48 < .001
control (-.01, -.01)
Immediate Nonverbal
gains visuospatial -.02
.09 -.02 -.19 .850
working (-.20, .17)
memory
Cognitive .03
.02 .11 1.12 .265
flexibility (-.02, .07)
-.26
Intercept .89 -.29 .770
(-2.03, 1.50)
Inhibitory -.01
.01 -.32 -3.34 .001
control (-.01, -.01)
Delayed Nonverbal
gains visuospatial .08
.12 .07 .70 .485
working (-.15, .32)
memory
Cognitive .03
.03 .09 .88 .379
flexibility (-.03, .09)
Note. R2 = .18 for the immediate gains. R2 = .12 for the delayed gains. p-values smaller
than .05 are highlighted in bold.

Finally, Tables 23 and 24 include the linear models of predictors in the picture-

description and article-noun congruent/incongruent tasks. All predictors failed to reach

statistical significance in both tasks.

 78

Table 23

Linear Model of Predictors of the Learning Gains of the Picture-Description Task with

95% Confidence Intervals (in parentheses)

Standard
Gains Predictor B β t-value p-value
error (B)
.62
Intercept .49 1.26 .212
(-.36, 1.59)
Inhibitory -.01
.01 -.11 -1.06 .291
control (-.01, .01)
Immediate Nonverbal
gains visuospatial .00
.07 .01 .01 .996
working (-.13, .13)
memory
Cognitive .01
.02 .03 .31 .757
flexibility (-.03, .04)
.93
Intercept .49 1.90 .060
(-.04, 1.89)
Inhibitory -.01
.01 -.20 -1.95 .054
control (-.01, .00)
Delayed Nonverbal
gains visuospatial -.02
.07 -.03 -.27 .785
working (-.15, .11)
memory
Cognitive .01
.02 .06 .57 .573
flexibility (-.02, .04)
Note. R2 = .01 for the immediate gains. R2 = .04 for the delayed gains.
 79

Table 24

Linear Model of Predictors of the Learning Gains of the Article-Noun

Congruent/Incongruent Task with 95% Confidence Intervals (in parentheses)

Standard
Gains Predictor B β t-value p-value
error (B)
17.29
Intercept 7.66 2.26 .026
(2.01, 32.50)
Inhibitory -.02
.02 -.09 -.89 .375
control (-.06, .02)
Immediate Nonverbal
gains visuospatial -1.10
1.02 -.11 -1.08 .283
working (-3.12, .92)
memory
Cognitive -.24
.26 -.10 -.91 .363
flexibility (-.75, .28)
17.29
Intercept 7.99 2.16 .033
(1.42, 33.16)
Inhibitory -.01
.02 -.03 -.29 .770
control (-.05, .03)
Delayed Nonverbal
gains visuospatial -1.43
1.06 -.14 -1.35 .181
working (-3.54, .68)
memory
Cognitive -.17
.27 -.07 -.64 .526
flexibility (-.71, .37)
Note. R2 = .02 for the immediate gains. R2 = .02 for the delayed gains. p-values smaller
than .05 are highlighted in bold.

4. 5. Summary

For the grammatical judgment task, students in both FFI conditions showed

significant improvement on the two posttests regardless of lexical familiarity and

frequency. They also demonstrated significantly higher scores for familiar but not

unfamiliar nouns in the text-completion task at the time of posttesting. Similarly, for the

article-noun congruent/incongruent task, students in both conditions attained significantly

higher scores on the posttests; in particular, those with FFI on both sublexical cues and
 80

pronunciation showed significantly higher scores for familiar nouns on the immediate

posttest.

In the forced-choice identification task, all participants, regardless of condition,

achieved maximum scores at all three testing times. For the picture-description task,

students receiving FFI on both sublexical cues and pronunciation attained significantly

higher scores on the posttests, favoring the familiar nouns. In addition, they also showed

significantly higher scores on the posttests in the read-aloud task. Whereas they showed

similar improvement in pronouncing un and une, they showed significantly higher scores

in pronouncing la compared to le. In contrast, students receiving FFI on only sublexical

cues did not demonstrate any significant improvement in the picture-description and

read-aloud tasks. As expected, the L1 baseline participants showed maximum scores on

all measures, and those in the control condition did not reveal any significant

improvement across the three testing sessions.

The scores of the grammatical judgment task and those of the read-aloud task

were significant predictors of students’ performance in the picture-description task on the

pretest. On the posttests, however, a different pattern emerged, with the scores of the

read-aloud task being the only predictor on the immediate and delayed posttests.

Finally, the learning gains made by students were mediated by their individual EF

skills. For instance, nonverbal visuospatial working memory was a significant predictor

of the gains in the grammatical judgment task. Inhibitory control was a significant

predictor of the gains in the read-aloud task.

 81

In the next chapter, I discuss the results of the current study focusing on the

differential effects of FFI on only sublexical cues and on both sublexical cues and

pronunciation as well as the roles of EF skills in L2 instruction.

 82

Chapter 5

Discussion

This chapter discusses the results of the present study. There are two sections in

the chapter. The first section addresses the differential effects of FFI on only sublexical

cues and on both sublexical cues and pronunciation on the acquisition of French

grammatical gender. The importance of L2 phonological knowledge in addition to L2

grammatical knowledge is highlighted in the first section. The second section addresses

the roles of EF skills in the acquisition of French grammatical gender. In particular, it

sheds light on the finding that the extent to which L2 learners benefit from instruction

depends on their EF skills. Finally, the chapter concludes with a summary that leads into

the concluding chapter.

5. 1. Differential Effects of Form-Focused Instruction on Only Sublexical Cues and

on Both Sublexical Cues and Pronunciation

The results of the present study revealed that students receiving FFI on only

sublexical cues showed significant improvement in the grammatical judgment, text-

completion, and article-noun congruent/incongruent tasks, but not in the read-aloud and

picture-description tasks. In contrast, those receiving FFI on both sublexical cues and

pronunciation showed significantly higher scores in all five tasks after the FFI sessions.

With respect to the forced-choice identification task, all participants obtained maximum

scores across the three testing sessions, indicating that they did not have any difficulty

perceptually categorizing the sounds un, une, le, and la either before or after the study.
 83

There were also effects of lexical familiarity. Students in both FFI conditions

showed significantly higher accuracy for familiar nouns, but not for unfamiliar nouns, in

the text-completion task at the time of posttesting. In a similar vein, those receiving FFI

on both sublexical cues and pronunciation attained significantly higher scores only for

familiar nouns in the picture-description task. In what follows, I discuss the differential

effects of FFI on only sublexical cues and on both sublexical cues and pronunciation by

task.

5. 1. 1. Grammatical Judgment and Text-Completion Tasks

In the grammatical judgment task, students in both FFI conditions showed

significant improvement after the FFI sessions. In particular, they obtained more than

90% accuracy for familiar, high-frequency unfamiliar, and low-frequency unfamiliar

nouns on the posttests. Both FFI conditions included several instructional techniques

drawing students’ attention to noun endings as predictors of gender attribution. For

instance, all target noun endings were highlighted in the texts along with their articles.

Students also completed a number of awareness activities (e.g., cloze, categorization, and

new word activities) in which they found various patterns in noun endings that predict

gender attribution and tested their knowledge with several nouns. Such activities induced

them to notice and internalize noun endings as reliable predictors of gender attribution

and thus develop more targetlike declarative knowledge regarding French grammatical

gender.

Students in both conditions also showed significant improvement in the text-

completion task. In addition to the aforementioned activities, those in both conditions

participated in writing activities in which they were incited to pay attention to the
 84

grammatical gender of given nouns in their compositions. It seems to be the case that the

writing activities predisposed them to retrieve their declarative knowledge and

proceduralize the declarative knowledge in the text-completion task (see skill acquisition

theory, DeKeyser, 1998, 2001; Lyster & Sato, 2013). The instructors in the present study

were also asked to employ various types of corrective feedback when they noticed any

errors in terms of gender attribution. Accordingly, corrective feedback might have helped

students to restructure their declarative knowledge towards greater accuracy and thus

facilitated more targetlike output.

However, the effects of FFI were limited to the familiar nouns in the text-

completion task. In contrast to the grammatical judgment task, students were required to

focus on not only sublexical cues but also other linguistic domains (e.g., L2 lexical,

syntactic, and pragmatic domains) to produce meaningful texts in French. The nature of

the task might have imposed more cognitive demands and detracted from the students’

attention to grammatical gender, particularly to noun endings to determine the gender

attribution of unfamiliar nouns. With respect to the familiar nouns, the students were

consistently prompted to retrieve the gender attribution of these nouns while engaging in

several activities during the FFI sessions. These practice opportunities might have led to

the proceduralization of the familiar nouns to some extent, which resulted in higher

accuracy for the familiar nouns in contrast to the unfamiliar nouns.

5. 1. 2. Forced-Choice Identification and Read-Aloud Tasks

In the forced-choice identification task, all participating students attained

maximum scores even on the pretest (i.e., more than 98.11%), meaning that they were

able to categorize the sounds un, une, le, and la in a targetlike manner before the study
 85

began. This finding was somewhat surprising since the current study predicted that L2

learners would have difficulty in both perceiving and producing the sounds, leading to

ambiguous pronunciation of French articles. In this regard, the finding confirmed that the

students in the present study were able to perceive the correct pronunciation of each

article and, thus, that their perception accuracy was not preventing them from showing

targetlike performance in oral production.

Notwithstanding their targetlike accuracy in the forced-choice identification task,

interesting enough, they had difficulty pronouncing the sounds of French articles un, une,

le, and la (in that order of difficulty) in the read-aloud task. As previous studies (e.g., Li

& Rosen, 2016) showed, they had most difficulty pronouncing un in a targetlike manner.

Such mismatches between their perception accuracy and production accuracy are

compatible with previous studies (Bradlow, Pisoni, Akahane-Yamada, & Tohkura, 1997;

Lee & Lyster, 2017). For instance, Lee and Lyster (2017) argued that L2 learners’

targetlike perception accuracy might be a requisite for targetlike L2 speech production,

but does not necessary result in targetlike L2 speech production. They also suggested that

explicit pronunciation instruction including ample production opportunities be considered

to facilitate targetlike L2 speech production. In this regard, the students receiving FFI on

both sublexical cues and pronunciation showed significant improvement in the read-aloud

task.

The articulation-based instruction offered at the beginning of the pronunciation

instruction provided them with metalinguistic information as to how each sound should

be articulated. They also participated in a number of production opportunities in which

they produced the target sounds, monitored their productions, and received corrective
 86

feedback including positive exemplars of the sounds. Accordingly, FFI on pronunciation

helped them articulate the sounds closer to targetlike norms. Owing to the lack of

pronunciation instruction, students receiving FFI on only sublexical cues were not able to

show any significant improvement in the read-aloud task.

5. 1. 3. Picture-Description and Article-Noun Congruent/Incongruent Tasks

In the picture-description task, students receiving FFI on both sublexical cues and

pronunciation showed significant improvement on the posttests. Due to the effects of FFI

drawing their attention to noun endings as predictors of gender attribution, as discussed in

the grammatical judgment and text-completion tasks, they drew on their knowledge of

noun endings to determine gender attribution. As reported in the read-aloud task, the

students had difficulty pronouncing the sounds of French articles at the time of pretesting,

and FFI on pronunciation helped them develop more targetlike pronunciation.

Consequently, those receiving FFI on both sublexical cues and pronunciation were able

to show higher performance on the posttests, thus having benefited from the combined

FFI not only to determine the gender of given nouns but also to pronounce the articles

more accurately.

However, its effects were limited to the familiar nouns. As in the text-completion

task, the students had to focus on other linguistic domains in the picture-description task.

For instance, they needed to consider other lexical items in addition to the target nouns

and L2 syntactic and phonological properties to describe pictures in spontaneous speech.

Consequently, the nature of the task imposed additional cognitive demands that prevented

participants from focusing on noun endings reliably to indicate the gender of the

unfamiliar nouns and thus from producing targetlike phrases in oral production. In
 87

addition, they had a number of opportunities to produce the familiar nouns with their

correct articles during the FFI sessions, which helped to proceduralize their use of

familiar nouns with correct articles. In the absence of similar practice opportunities with

the unfamiliar nouns, participants fared less well in qualifying them with correct gender

markers.

As discussed in regard to the grammatical judgment and text-completion tasks,

the students receiving FFI on only sublexical cues were able to draw on their knowledge

of noun endings to determine gender attribution as much as those receiving FFI on both

sublexical cues and pronunciation. Nevertheless, the former students failed to show any

improvement in the picture-description task. Considering that the students in the current

study had difficulty articulating the sounds of French articles, presumably their lack of L2

phonological knowledge prevented them from showing targetlike performance in spite of

their accurate grammatical knowledge.

The importance of L2 phonological knowledge is also supported by the multiple

regression analyses. On the pretest, the scores of the grammatical judgment task and

those of the read-aloud task were significant predictors of students’ performance in the

picture-description task. In other words, their nontargetlike performance in the picture-

description task resulted from not only their lack of grammatical knowledge in gender

attribution but also their lack of phonological knowledge in producing the sounds of

French articles. Its importance was even more evident after the FFI sessions. On the

posttests, the scores on the read-aloud task were the only significant predictor of the

scores on the picture-description task. As such, the results of the current study indicate
 88

that targetlike L2 phonological knowledge is needed for L2 learners to show targetlike

performance in oral production.

In the article-noun congruent/incongruent task, students in both FFI conditions

showed significant improvement. They showed higher scores for the familiar and

unfamiliar nouns on the posttests. Given that they did not have any difficulty perceiving

un, une, le, and la in the forced-choice identification task, their performance in the

article-noun congruent/incongruent task was more likely to depend on their accuracy in

gender attribution. Due to the effects of FFI on sublexical cues, the students in both FFI

conditions were able to increase their accuracy in assigning grammatical gender, which in

turn helped them to improve significantly in the article-noun congruent/incongruent task

after the FFI sessions. One of the noteworthy points is that the stimuli (i.e., the target

nouns) were provided aurally in this task. The instructional components in the FFI

explicitly induced the students to pay attention to noun endings focusing on their

orthographic representations (e.g., ‘-eau’ → masculine). Since all target nouns appeared

visually in written form in the grammatical judgment and text-completion tasks, it may

have relatively been easy for students to draw on their knowledge of noun endings to

assign grammatical gender. In the article-noun congruent/incongruent task, on the

contrary, they were required to rely on the phonemic information of given nouns to

determine their gender attribution. The finding that participants’ accuracy increased for

both unfamiliar nouns and familiar nouns on the posttests suggests two possible ways of

processing gender cues. The first way is for participants, when hearing the noun chapeau

/ʃapo/, to focus on the final phoneme /o/ and then search for possible noun endings

representing the phoneme /o/ (e.g., -eau, -ot, or -o), after which they could assign its
 89

grammatical gender based on the orthographic representations. The second way is for

participants to encode the orthographic and phonemic representations at the same time.

For instance, during the instructional activities, the instructors consistently verbalized the

target nouns while drawing the students’ attention to noun endings to indicate gender

attribution. Therefore, the students could encode both representations (e.g., ‘-eau’ - /o/)

with gender information (e.g., masculine) and retrieve both simultaneously to determine

gender attribution when hearing /ʃapo/. In either case, the current study lends support to

Hardison’s (1992) argument that L2 learners exploit both the phonemic and orthographic

representations of noun endings to predict gender attribution.

5. 1. 4. Summary

In sum, FFI on sublexical cues was beneficial for L2 learners to improve their

accuracy of French grammatical gender by drawing their attention to noun endings as

predictors of gender attribution. These findings are also compatible with those of

previous studies (Harley, 1998, Lyster, 2004; Lyster & Izquierdo, 2009; Warden, 1997).

Although the students in the control condition were given a list of noun endings (only 10

nouns) with their predictability in gender attribution, they failed to show any significant

improvement on any measures. Accordingly, L2 learners need FFI and ample practice

opportunities rather than being given a list of noun endings and minimal exercises.

More importantly, pronunciation instruction played a complementary role in the

acquisition of French grammatical gender. It enabled L2 learners to develop more

targetlike pronunciation of the sounds un, une, le and la, which helped them approximate

targetlike norms in oral production (e.g., picture-description task). In this regard, the
 90

current study suggests that pronunciation instruction be included in L2 instruction

targeting French grammatical gender.

5. 2. Roles of Executive Function Skills in Second Language Development

The present study found that nonverbal and nonlinguistic EF skills mediated the

extent to which the students benefit from FFI. There were task-specific effects.

Nonverbal visuospatial working memory was a significant predictor of the learning gains

in the grammatical judgment task. Inhibitory control was the only significant predictor of

the learning gains in the read-aloud task. Cognitive flexibility failed to predict any gains.

FFI pushed the students to focus on noun endings to determine gender attribution.

To do so, the students had to pay attention to the orthographic representations of target

noun endings (e.g., ‘-ent’, ‘-tion’, or ‘-eau’), which results in orthographic processing.

According to Pham and Hasson (2014), orthographic processing entails attention to “the

formation of visual representations of letters, letter patterns, and sequences of letters that

serve to map spatially the temporal sequence of phonemes within words” (p. 474). In this

regard, those having high nonverbal visuospatial working memory might be at an

advantage in detecting, encoding, and retrieving the orthographic representations of noun

endings to assign grammatical gender. Considering that the grammatical judgment task

exclusively required them to detect noun endings and then determine gender attribution in

a controlled setting, the role of nonverbal visuospatial working memory seems to be

evident in this specific task.

Previous studies (Darcy, Mora, & Daidone, 2014; Darcy et al., 2016) found that

inhibitory control is important for targetlike L2 speech perception and production of L2

 91

segments. In a similar vein, the current study revealed that inhibitory control was a

significant predictor of the degree to which the students improved their production

accuracy of the sounds un, une, le, and la. As argued by Darcy et al. (2016), those

showing high inhibitory control might have been able to articulate the L2 sounds in a

targetlike manner while inhibiting their L1 phonology and tuning into phonologically

relevant acoustic properties in French.

None of the students’ L1s had grammatical gender. The current study predicted

that those with high cognitive flexibility would be more aware of grammatical gender,

which is a missing feature in their L1s, and benefit more from FFI by having rapid and

accurate switching between their gender-neutral L1s and L2 French. However, this

variable failed to predict any gains. Given previous studies (e.g., Kapa & Colombo, 2014;

Stone & Pili-Moss, 2015) showing mixed results, the role of cognitive flexibility in L2

learning is somewhat uncertain and needs more empirical testing.

The results of the present study lead to the conclusion that some EF skills (i.e.,

nonverbal visuospatial working memory and inhibitory control) are important in L2

learning and instruction. The gains made in the grammatical judgment task were affected

by nonverbal visuospatial working memory, while the gains made in the read-aloud task

were influenced by inhibitory control. As such, the effects of EF skills were found only in

the relatively controlled tasks—the grammatical judgment and read-aloud tasks.

Considering that the picture-description and text-completion tasks required L2 learners to

draw on multiple aspects of their L2 knowledge (e.g., L2 lexical, syntactic, and

phonological knowledge), the impact of the EF skills proved less evident in these tasks

than in the tasks requiring knowledge of a single domain.

 92

5. 3. Summary

This chapter discussed the results of the current study. FFI on sublexical cues was

effective for the students to increase their accuracy in assigning grammatical gender.

Moreover, pronunciation instruction contributed even more to the acquisition of French

grammatical gender by enabling the students to develop targetlike pronunciation of the

sounds.

Participants’ learning gains were mediated by their EF skills. For instance, those

with high nonverbal visuospatial working memory had an advantage in terms of

detecting, encoding, and retrieving noun endings as predictors of gender attribution. In

addition, participants with high inhibitory control were better able to articulate the sounds

un, une, le, and la in a targetlike manner as they focused on phonologically relevant

acoustic properties in the target language without interference from their L1 phonology.

In the next chapter, I conclude the present study by summarizing it and proposing

pedagogical implications in addition to future directions.

 93

Chapter 6

Conclusion

This chapter concludes the dissertation. After a summary of the current study, its

pedagogical implications are addressed in relation to the field of L2 education. Finally,

future directions are then proposed in light of the limitations of the current study.

6. 1. Summary of the Current Study

In contrast to L1 speakers, French L2 learners are known to have difficulty

acquiring grammatical gender in spite of the frequency of gender markers in linguistic

input (Harley, 1998; Lyster, 2004). Considering that L2 learners’ errors in grammatical

gender are likely to occur as a result of their lack of knowledge of gender attribution

(Grüter et al., 2012) and that noun endings are important predictors of gender attribution

in French (Lyster, 2006), the current study attempted to investigate the extent to which

L2 learners benefit from FFI drawing their attention to noun endings as a means of

determining gender attribution.

One of the interesting reports from previous studies (Harley, 1998; Lyster, 2004)

is that L2 learners tend to produce ambiguous pronunciation of French articles that leads

to a lack of intelligibility and grammatical inaccuracy after all. Considering that L2

learners have difficulty acquiring /œ̃/ (in un) and /y/ (in une) (Li & Rosen, 2016), the

present study hypothesized that L2 learners’ phonological difficulty impeded them from

acquiring French grammatical gender in a targetlike manner. As such, the current study

also tested FFI on pronunciation to examine whether it could facilitate their acquisition of

French grammatical gender in oral production. In addition, given that L2 learners’ EF

 94

skills are important in L2 learning (Darcy et al., 2016; Kapa & Colombo, 2014; Linck et

al., 2014), the current study investigated whether the degree to which L2 learners benefit

from FFI would be mediated by their EF skills, which, in the present study, included

inhibitory control, nonverbal visuospatial working memory, and cognitive flexibility.

A quasi-experimental study was conducted with a total of 140 L2 students in

Elementary French 1 (FRSL 207/208) at McGill University. Two instructional conditions

(i.e., FFI on only sublexical cues and FFI on both sublexical cues and pronunciation) in

addition to one control condition were implemented in six different classrooms (two

classrooms per condition). Those in the two FFI conditions received six 80-minute

instructional sessions. The students in the first FFI condition partook in a number of

instructional activities drawing their attention to noun endings as predictors of gender

attribution, whereas those in the second FFI condition received not only FFI on sublexical

cues but also pronunciation instruction targeting the sounds of French articles, un, une, le,

and la. In order to equalize instructional hours between the two FFI conditions, the

former condition had more meaning-focused activities. The students in the control

condition received regular L2 French lessons excluding any FFI components adopted in

the two FFI conditions. A total of six linguistic tasks, measuring their grammatical

knowledge in gender attribution and their phonological knowledge in producing and

perceiving the sounds of French articles, were conducted at three different times. The

Simon Test (inhibitory control), the Corsi Block-Tapping Test (nonverbal visuospatial

working memory), and the Wisconsin Card Sorting Test (cognitive flexibility) were also

administered to measure L2 students’ EF skills.

 95

Results showed that FFI on sublexical cues is beneficial for the students to

increase their accuracy in determining gender attribution. More importantly, FFI on

pronunciation contributed to the acquisition of French grammatical gender by enabling

the students to develop more targetlike pronunciation of the sounds and to show better

performance in oral production (e.g., the picture-description task). In addition, students’

learning gains were mediated by their EF skills, specifically nonverbal visuospatial

working memory in the grammatical judgment task and inhibitory control in the read-

aloud task.

The present study confirmed the effects of L2 phonological knowledge and

pronunciation instruction on the acquisition of French grammatical gender. It also

demonstrated the roles of EF skills in L2 acquisition. Based on these findings, I elaborate

pedagogical implications as follows.

6. 2. Pedagogical Implications

The findings in the present study lead to several pedagogical implications. First,

the current study showed that FFI on sublexical cues is beneficial for L2 learners to

notice and internalize noun endings as predictors of gender attribution. In line with

previous studies (Harley, 1998, Lyster, 2004; Lyster & Izquierdo, 2009; Warden, 1997),

it is recommended for L2 practitioners to consider adapting FFI on sublexical cues

including noticing, awareness, and practice activities instead of either teaching

grammatical gender on an item-by-item basis or expecting their students to learn it

incidentally. The instructors in the control condition provided their students with some

sets of noun endings and briefly explained their predictability in gender attribution.
 96

Nevertheless, those in the control condition did not show any significant improvement on

any measures. In this regard, L2 practitioners need to help their students to find the

patterns by themselves and to practice them through controlled and spontaneous practice

activities. Based on skill acquisition theory (DeKeyser, 1998, 2001; Lyster & Sato,

2013), such practice opportunities help students to proceduralize declarative knowledge

in a targetlike manner.

The present study also found that L2 phonological knowledge plays an important

role in the acquisition of French grammatical gender. Targetlike L2 phonological

knowledge helps L2 learners to show better performance in oral production. In this sense,

L2 practitioners should consider implementing pronunciation instruction focusing on the

sounds of French articles along with FFI on sublexical cues. In line with previous studies

(Lee & Lyster, 2017; Saito, 2013), explicit phonetic instruction including articulatory

information as to how each sound is articulated is highly recommended. As contended by

Lee and Lyster (2017), ample practice opportunities should be made available for L2

learners to develop targetlike production accuracy.

6. 3. Future Directions

In this section, I propose future directions drawing on the limitations of the

current study. The students in the present study spoke several L1s and learned other L2s

in addition to French. Although all students whose L1 had grammatical gender were

removed, there were a number of students who learned other L2s with grammatical

gender (e.g., Spanish), which could have affected their acquisition of French grammatical

gender. In addition, Mandarin and Turkish have /y/ as a separate phoneme, which might
 97

help Mandarin L1 and Turkish L1 speakers produce the sound une with less difficulty in

contrast to other students. Accordingly, for future studies, it would be interesting to see if

L1s and other L2s affect the acquisition of French grammatical gender and the

effectiveness of FFI (and, if so, to what extent?) by controlling for students’ L1s and

other L2s.

The effects of FFI were limited to the familiar nouns in the text-completion and

picture-description tasks. It was speculated in the discussion that this finding was due to

the nature of tasks requiring participants to use multiple linguistic domains such as L2

lexical, syntactic, and pragmatic domains. Given that these two tasks are similar to those

in real-life language use, it would be pedagogically important to delve into how L2

practitioners can help L2 learners increase their accuracy with unfamiliar nouns.

FFI in the current study drew L2 students’ attention to noun endings, and those in

both FFI conditions showed significant improvement for familiar, high-frequency

unfamiliar, and low-frequency unfamiliar nouns in the grammatical judgment task. As

such, it was inferred that the students showed higher accuracy after the FFI sessions since

they focused on noun endings to predict gender attribution. To examine the extent to

which L2 learners in fact use noun endings to determine gender attribution in the task, it

would be intriguing to employ other procedures such as eye-tracking techniques.

In the present study, the Simon Test, the Corsi Block-Tapping Test, and the

Wisconsin Card Sorting Test were adopted to measure the students’ EF skills. The

present study found partial effects of EF skills on the learning gains. It would be worth

considering other EF tests to have a broader understanding of EF skills in L2 learning and

teaching.
 98

Given the roles of EF skills in L2 learning, Kapa and Colombo (2014) stated that

“it may be possible to integrate EF training, which is becoming increasingly popular (...),

along with language instruction in order to improve individuals’ language learning

outcomes” (p. 250). Numerous studies (e.g., Diamond, Barnett, Thomas, & Munro, 2007;

Diamond & Lee, 2011; Thorell, Lindqvist, Bergman Nutley, Bohlin, & Klingberg, 2009)

showed the effectiveness of training on EF skills and possible transfer effects between EF

skills. Blair and Razza (2007) and Diamond and Lee (2011) found that training on EF

skills facilitates school readiness and academic success such as math skills and literacy

levels. In this sense, it would also be interesting to investigate whether L2 learners with

training on EF skills would be at an advantage in L2 acquisition and processing in

contrast to L2 learners without training. I believe that this research question will yield

important implications in the field of L2 education and acquisition.

Finally, a number of previous studies targeting L2 pronunciation domains have

found that L2 pronunciation instruction is effective for L2 learners to improve their

accuracy in L2 speech production and perception (Lee & Lyster, 2016a, 2016b, 2017;

Saito, 2013; Saito & Lyster, 2012; Saito & Wu, 2014). The current study showed new

insights into L2 pronunciation instruction; that is, the importance of L2 phonological

knowledge and pronunciation instruction in L2 grammar teaching and learning focusing

on French grammatical gender. It is hoped that future studies investigate the role of L2

phonological knowledge in L2 grammar and vocabulary learning in a broader sense, as

well as the effects of L2 pronunciation instruction on the acquisition of L2 lexical and

morphological targets.
 99

References

Atkins, P. W., & Baddeley, A. (1998). Working memory and distributed vocabulary

learning. Applied Psycholinguistics, 19(4), 537-552.

Baddeley, A. (2000). The episodic buffer: A new component of working memory?

Trends in Cognitive Sciences, 4(11), 417-423.

Baddeley, A. (2003). Working memory: Looking back and looking forward. Nature

Reviews Neuroscience, 4(10), 829-839.

Baddeley, A. D., & Hitch, G. (1974). Working memory. Psychology of Learning and

Motivation, 8, 47-89.

Bartning, I. (2000). Gender agreement in L2 French: Pre‐advanced vs advanced learners.

Studia Linguistica, 54(2), 225-237.

Bernstein, J. B. (1993). Topics in the syntax of nominal structure across Romance.

(Unpublished doctoral dissertation), City University of New York, Graduate

Center, New York, USA.

Bialystok, E., Craik, F. I., Klein, R., & Viswanathan, M. (2004). Bilingualism, aging, and

cognitive control: Evidence from the Simon task. Psychology and Aging, 19(2),

290-303.

Bialystok, E., & Viswanathan, M. (2009). Components of executive control with

advantages for bilingual children in two cultures. Cognition, 112(3), 494-500.

Blair, C., & Razza, R. P. (2007). Relating effortful control, executive function, and false

belief understanding to emerging math and literacy ability in kindergarten. Child

Development, 78(2), 647-663.

 100

Bradlow, A. R., Pisoni, D. B., Akahane-Yamada, R., & Tohkura, Y. (1997). Training

Japanese listeners to identify English /r/ and /l/: IV. Some effects of perceptual

learning on speech production. Journal of the Acoustical Society of America,

101(4), 2299-2310.

Caramazza, A. (1997). How many levels of processing are there in lexical access?

Cognitive Neuropsychology, 14(1), 177-208.

Caramazza, A., & Miozzo, M. (1997). The relation between syntactic and phonological

knowledge in lexical access: Evidence from the ‘tip-of-the-tongue’ phenomenon.

Cognition, 64(3), 309-343.

Carroll, S. (1989). Second-language acquisition and the computational paradigm.

Language Learning, 39(4), 535-594.

Carstens, V. (2000). Concord in minimalist theory. Linguistic Inquiry, 31(2), 319-355.

Chen, Y. (1999). Les lettres chinoises. Montreal, QC: Leméac.

Chomsky, N. (1995). The minimalist program. Cambridge, MA: MIT Press.

Chun, D. M., & Payne, J. S. (2004). What makes students click: Working memory and

look-up behavior. System, 32(4), 481-503.

Clark, E. V. (1985). The acquisition of Romance, with special reference to French. In D.

I. Slobin (Ed.), The crosslinguistic study of language acquisition (Vol. 1, pp. 687-

782). Hillsdale, NJ: Lawrence Erlbaum.

Corsi, P. M. (1972). Human memory and the medial temporal region of the brain.

(Unpublished doctoral dissertation), McGill University, Montreal, QC.

 101

Costa, A., & Santesteban, M. (2004). Lexical access in bilingual speech production:

Evidence from language switching in highly proficient bilinguals and L2 learners.

Journal of Memory and Language, 50(4), 491-511.

Cubelli, R., Lotto, L., Paolieri, D., Girelli, M., & Job, R. (2005). Grammatical gender is

selected in bare noun production: Evidence from the picture–word interference

paradigm. Journal of Memory and Language, 53(1), 42-59.

Darcy, I., Mora, J. C., & Daidone, D. (2014). Attention control and inhibition influence

phonological development in a second language. Concordia Working Papers in

Applied Linguistics, 5, 115-129.

Darcy, I., Mora, J. C., & Daidone, D. (2016). The role of inhibitory control in second

language phonological processing. Language Learning, 66(4), 741-773.

DeKeyser, R. M. (1998). Beyond focus on form: Cognitive perspectives on learning and

practicing second language grammar. In C. J. Doughty & J. Williams (Eds.),

Focus on form in classroom second language acquisition (pp. 42-63). Cambridge:

Cambridge University Press.

DeKeyser, R. M. (2001). Automaticity and automatization. In P. Robinson (Ed.),

Cognition and second language instruction (pp. 125-151). Cambridge: Cambridge

University Press.

Desrochers, A., Paivio, A., & Desrochers, S. (1989). L’effet de la fréquence d’usage des

noms inanimés et de la valeur prédictive de leur terminaison sur l’identification

du genre grammatical. Canadian Journal of Psychology/Revue Canadienne de

Psychologie, 43(1), 62-73.

 102

Diamond, A. (2012). Activities and programs that improve children’s executive

functions. Current Directions in Psychological Science, 21(5), 335-341.

Diamond, A., Barnett, W. S., Thomas, J., & Munro, S. (2007). Preschool program

improves cognitive control. Science, 318(5855), 1387-1388.

Diamond, A., & Lee, K. (2011). Interventions shown to aid executive function

development in children 4 to 12 years old. Science, 333(6045), 959-964.

French, L. M., & O’Brien, I. (2008). Phonological memory and children’s second

language grammar learning. Applied Psycholinguistics, 29(3), 463-487.

Gangopadhyay, I., Davidson, M. M., Weismer, S. E., & Kaushanskaya, M. (2016). The

role of nonverbal working memory in morphosyntactic processing by school-aged

monolingual and bilingual children. Journal of Experimental Child Psychology,

142, 171-194.

Goad, H., & White, L. (2006). Ultimate attainment in interlanguage grammars: A

prosodic approach. Second Language Research, 22(3), 243-268.

Goad, H., White, L., & Steele, J. (2003). Missing inflection in L2 acquisition: Defective

syntax or L1-constrained prosodic representations? Canadian Journal of

Linguistics/La Revue Canadienne de Linguistique, 48(3/4), 243-263.

Gooch, R., Saito, K., & Lyster, R. (2016). Effects of recasts and prompts on L2

pronunciation development: Teaching English /ɹ/ to Korean adult EFL learners.

System, 60, 117-127.

Grant, D. A., & Berg, E. (1948). A behavioral analysis of degree of reinforcement and

ease of shifting to new responses in a Weigl-type card-sorting problem. Journal of

Experimental Psychology, 38(4), 404-411.

 103

Green, D. W. (1998). Mental control of the bilingual lexico-semantic system.

Bilingualism: Language and Cognition, 1(2), 67-81.

Grüter, T., Lew-Williams, C., & Fernald, A. (2012). Grammatical gender in L2: A

production or a real-time processing problem? Second Language Research, 28(2),

191-215.

Hardison, D. M. (1992). Acquisition of grammatical gender in French: L2 learner

accuracy and strategies. Canadian Modern Language Review/La Revue

Canadienne des Langues Vivantes, 48(2), 292-306.

Harley, B. (1979). French gender “rules” in the speech of English-dominant, French-

dominant and monolingual French-speaking children. Working Papers on

Bilingualism, 19, 129-156.

Harley, B. (1998). The role of form-focused tasks in promoting child L2 acquisition. In

C. J. Doughty & J. Williams (Eds.), Focus on form in classroom second language

acquisition (pp. 156-174). Cambridge: Cambridge University Press.

Higgins, E. (2009). Libre opinion : L’anglo de Saint Pierre. Le Devoir. Retrieved from

[Link]

Holmes, V. M., & Dejean de la Bâtie, B. (1999). Assignment of grammatical gender by

native speakers and foreign learners of French. Applied Psycholinguistics, 20(4),

479-506.

Juffs, A. (2004). Representation, processing and working memory in a second language.

Transactions of the Philological Society, 102(2), 199-225.

Kapa, L. L., & Colombo, J. (2014). Executive function predicts artificial language

learning. Journal of Memory and Language, 76, 237-252.

 104

Karmiloff-Smith, A. (1979). A functional approach to child language. Cambridge:

Cambridge University Press.

Kok, A. (1999). Varieties of inhibition: Manifestations in cognition, event-related

potentials and aging. Acta Psychologica, 101(2-3), 129-158.

La Heij, W., Mak, P., Sander, J., & Willeboordse, E. (1998). The gender-congruency

effect in picture-word tasks. Psychological Research, 61(3), 209-219.

Larson-Hall, J. (2010). A guide to doing statistics in second language research using

SPSS. New York, NY: Routledge.

Laufer, B., & Girsai, N. (2008). Form-focused instruction in second language vocabulary

learning: A case for contrastive analysis and translation. Applied Linguistics,

29(4), 694-716.

Laurin, J., & Jacob, R. (2006). Ma grammaire (Nouvelle édition ed.). Montreal, QC: Les

Éditions de l’Homme.

Lee, A. H., & Lyster, R. (2016a). The effects of corrective feedback on instructed L2

speech perception. Studies in Second Language Acquisition, 38(1), 35-64.

Lee, A. H., & Lyster, R. (2016b). Effects of different types of corrective feedback on

receptive skills in a second language: A speech perception training study.

Language Learning, 66(4), 809-833.

Lee, A. H., & Lyster, R. (2017). Can corrective feedback on second language speech

perception errors affect production accuracy? Applied Psycholinguistics, 38(2),

371-393.

Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in

speech production. Behavioral and Brain Sciences, 22(1), 1-38.

 105

Li, F., & Rosen, N. (2016). Vowel production development of French-immersion students

in Southern Alberta. Poster presented at the annual meeting of the Canadian

Linguistic Association, Calgary, AB.

Linck, J. A., Osthus, P., Koeth, J. T., & Bunting, M. F. (2014). Working memory and

second language comprehension and production: A meta-analysis. Psychonomic

Bulletin & Review, 21(4), 861-883.

Lissouba, B. D. (1994). Les libres propos de Binéka. Saint Maur: Sépia.

Lyster, R. (2004). Differential effects of prompts and recasts in form-focused instruction.

Studies in Second Language Acquisition, 26(3), 399-432.

Lyster, R. (2006). Predictability in French gender attribution: A corpus analysis. Journal

of French Language Studies, 16(1), 69-92.

Lyster, R., & Izquierdo, J. (2009). Prompts versus recasts in dyadic interaction. Language

Learning, 59(2), 453-498.

Lyster, R., & Sato, M. (2013). Skill acquisition theory and the role of practice in L2

development. In P. García Mayo, M. Gutierrez-Mangado, & M. Martínez Adrián

(Eds.), Contemporary approaches to second language acquisition (pp. 71-92).

Amsterdam: John Benjamins.

Martin-Rhee, M. M., & Bialystok, E. (2008). The development of two types of inhibitory

control in monolingual and bilingual children. Bilingualism: Language and

Cognition, 11(1), 81-93.

 106

McDonough, K., & Kim, Y. (2016). Working memory and L2 English speakers’ primed

and subsequent production of passives. In G. Granena, D. O. Jackson, & Y.

Yilmaz (Eds.), Cognitive individual differences in second language processing

and acquisition (pp. 205-222). Amsterdam: John Benjamins.

McDonough, K., & Trofimovich, P. (2016). The role of statistical learning and working

memory in L2 speakers’ pattern learning. Modern Language Journal, 100(2),

428-445.

Milner, B. (1971). Interhemispheric differences in the localization of psychological

processes in man. British Medical Bulletin, 27(3), 272-277.

Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T.

D. (2000). The unity and diversity of executive functions and their contributions

to complex “frontal lobe” tasks: A latent variable analysis. Cognitive Psychology,

41(1), 49-100.

Miyake, A., & Shah, P. (1999). Models of working memory: Mechanisms of active

maintenance and executive control. Cambridge: Cambridge University Press.

New, B., Pallier, C., Ferrand, L., & Matos, R. (2001). Une base de données lexicales du

français contemporain sur internet: LEXIQUE™. L’Année Psychologique, 101(3),

447-462.

Nguyen, T. T. M., Pham, T. H., & Pham, M. T. (2012). The relative effects of explicit

and implicit form-focused instruction on the development of L2 pragmatic

competence. Journal of Pragmatics, 44(4), 416-434.

 107

Nicolay, A.-C., & Poncelet, M. (2013). Cognitive advantage in children enrolled in a

second-language immersion elementary school program for three years.

Bilingualism: Language and Cognition, 16(3), 597-607.

Nigg, J. T. (2000). On inhibition/disinhibition in developmental psychopathology: Views

from cognitive and personality psychology and a working inhibition taxonomy.

Psychological Bulletin, 126(2), 220-246.

Paradis, J., & Prévost, P. (2004). Functional categories in the acquisition of French. In J.

Paradis & P. Prévost (Eds.), The acquisition of French in different contexts:

Focus on functional categories (pp. 1-23). Amsterdam: John Benjamins.

Pham, A. V., & Hasson, R. M. (2014). Verbal and visuospatial working memory as

predictors of children’s reading ability. Archives of Clinical Neuropsychology,

29(5), 467-477.

Picallo, M. C. (1991). Nominals and nominalizations in Catalan. Probus, 3(3), 279-316.

Poarch, G. J., & van Hell, J. G. (2012). Executive functions and inhibitory control in

multilingual children: Evidence from second-language learners, bilinguals, and

trilinguals. Journal of Experimental Child Psychology, 113(4), 535-551.

R Core Team. (2016). R: A language and environment for statistical computing. Vienna,

Austria: R Foundation for Statistical Computing. Retrieved from [Link]

[Link]/

Ranta, L., & Lyster, R. (2018). Form-focused instruction. In P. Garrett & J. Cots (Eds.),

The Routledge handbook of language awareness (pp. 40-56). New York, NY:

Routledge.
 108

Ritter, E. (1991). Two functional categories in noun phrases: Evidence from modern

Hebrew. In S. D. Rothstein (Ed.), Syntax and semantics (Vol. 25. Perspectives on

phrase structure: Heads and licensing, pp. 37-62). Bingley: Emerald.

Roelofs, A. (1992). A spreading-activation theory of lemma retrieval in speaking.

Cognition, 42(1-3), 107-142.

Saito, K. (2013). Re-examining effects of form-focused instruction on L2 pronunciation

development: The role of explicit phonetic information. Studies in Second

Language Acquisition, 35(1), 1-29.

Saito, K., & Lyster, R. (2012). Effects of form-focused instruction and corrective

feedback on L2 pronunciation development of /ɹ/ by Japanese learners of English.

Language Learning, 62(2), 595-633.

Saito, K., & Wu, X. (2014). Communicative focus on form and L2 suprasegmental

learning: Teaching Cantonese learners to perceive Mandarin tones. Studies in

Second Language Acquisition, 36(4), 647-680.

Schmidt, R. (2001). Attention. In P. Robinson (Ed.), Cognition and second language

instruction (pp. 3-32). Cambridge: Cambridge University Press.

Seçer, I. (2016). Skills of cognitive flexibility in monolingual and bilingual younger

adults. Journal of General Psychology, 143(3), 172-184.

Sera, M. D., Elieff, C., Forbes, J., Burch, M. C., Rodríguez, W., & Dubois, D. P. (2002).

When language affects cognition and when it does not: An analysis of

grammatical gender and classification. Journal of Experimental Psychology:

General, 131(3), 377-397.

 109

Sharwood Smith, M. (1993). Input enhancement in instructed SLA: Theoretical bases.

Studies in Second Language Acquisition, 15(2), 165-179.

Simon, J. R., & Rudell, A. P. (1967). Auditory S-R compatibility: The effect of an

irrelevant cue on information processing. Journal of Applied Psychology, 51(3),

300-304.

Sokolik, M. E., & Smith, M. E. (1992). Assignment of gender to French nouns in primary

and secondary language: A connectionist model. Second Language Research,

8(1), 39-58.

Spada, N. (1997). Form-focused instruction and second language acquisition: A review of

classroom and laboratory research. Language Teaching, 30(2), 73-87.

Spada, N., Jessop, L., Tomita, Y., Suzuki, W., & Valeo, A. (2014). Isolated and

integrated form-focused instruction: Effects on different types of L2 knowledge.

Language Teaching Research, 18(4), 453-473.

Stoet, G. (2010). PsyToolkit: A software package for programming psychological

experiments using Linux. Behavior Research Methods, 42(4), 1096-1104.

Stoet, G. (2017). PsyToolkit: A novel web-based method for running online

questionnaires and reaction-time experiments. Teaching of Psychology, 44(1), 24-

31.

Stone, H., & Pili-Moss, D. (2015). Executive function and language learning:

Differentiating vocabulary and morpho-syntax. In E. Nichele, D. Pili-Moss, & C.

Sitthirak (Eds.), Papers from the 10th Lancaster University Postgraduate

Conference in Linguistics & Language Teaching 2015 (Vol. 10, pp. 53-74).

Lancashire: Lancaster University.

 110

Surridge, M. E., & Lessard, G. (1984). Pour une prise de conscience du genre

grammatical (towards a grasp of grammatical gender). Canadian Modern

Language Review/La Revue Canadienne des Langues Vivantes, 41(1), 43-52.

Swain, M. (1985). Communicative competence: Some roles of comprehensible input and

comprehensible output in its development. In S. M. Gass & C. Madden (Eds.),

Input in second language acquisition (pp. 235-253). Rowley, MA: Newbury

House.

Swain, M. (1995). Three functions of output in second language learning. In G. Cook &

B. Seidlhofer (Eds.), Principle and practice in applied linguistics: Studies in

honour of H. G. Widdowson (pp. 125-144). Oxford: Oxford University Press.

Thorell, L. B., Lindqvist, S., Bergman Nutley, S., Bohlin, G., & Klingberg, T. (2009).

Training and transfer effects of executive functions in preschool children.

Developmental Science, 12(1), 106-113.

Tucker, G. R., Lambert, W. E., & Rigault, A. (1977). The French speaker’s skill with

grammatical gender: An example of rule-governed behavior. Paris: Mouton De

Gruyter.

Tucker, G. R., Lambert, W. E., Rigault, A., & Segalowitz, N. (1968). A psychological

investigation of French speakers’ skill with grammatical gender. Journal of

Verbal Learning and Verbal Behavior, 7(2), 312-316.

van Heugten, M., & Shi, R. (2009). French-learning toddlers use gender information on

determiners during word recognition. Developmental Science, 12(3), 419-425.

 111

Warden, M. (1997). The effect of form-focused instruction on control over grammatical

gender by French immersion students in grade 11. (Unpublished doctoral

dissertation), University of Toronto, Ontario, Canada.

 112

Appendix A

Informed Consent Form for Participating Students

INFORMED CONSENT FORM FOR PARTICIPATING FRENCH LEARNERS

(FRENCH L2 PARTICIPANTS)
Dear Student,

Thank you for your interest in participating in our SSHRC-funded research study (# 430-
2017-00372). The purpose of this study is to investigate the extent to which learners of
French as a second language benefit from different types of instruction on grammatical
gender (REB file # 93-0817). In order to achieve the goal of the current study, your
participation will be highly valued.

Your classroom has been chosen to collaborate in joint project with the Faculty of
Education. This project involves piloting some new instructional techniques that aim to
increase your ability to accurately use grammatical gender in French. Your instructor will
integrate a focus on grammatical gender for a total of about 8 hours (six 80-minute
instructional sessions) during the first half of the semester into the activities that normally
comprise FRSL 207/208. In addition, on three different occasions, you will partake in
computer-assisted tasks targeting grammatical gender as extra practice during lab
sessions. The purpose of this letter is to ask for your permission to use the results of these
assessment tasks for research purposes.

Here is a description of the tasks you will complete. At Time 1, Time 2, and Time 3, you
will (a) identify correct French articles (e.g., un or une for the word stylo); (b) write
compositions with given words; (c) identify whether a given word, played via a headset,
is un, une, le, or la; (d) read French words; (e) describe pictures; and (f) identify whether
a given phrase, played via a headset, is grammatical. Also at each of the three testing
times, your executive function skills will be measured.

Your utterances will be audio recorded and then rated by native speakers of French. Due
to the recognizability of the human voice, the confidentiality of your identity will not be
completely guaranteed. Nevertheless, your personal information will not be publicized
under any circumstances and your speech will be solely used for this study.

The results obtained from the above tests will not affect your grade in the course; your
instructor will not know whether or not you agreed to let us use the test results, so you are
under no pressure or obligation to give us consent. Your participation is completely
voluntary. Even if you agree now to participate, you can change your mind later and
withdraw from the study at any time without any negative consequence.
 113

There is monetary compensation of $80 for releasing your results regardless of whether
you complete all duties and/or withdraw from the study.

The results of this study will be submitted for peer review and publication in professional
journals, newsletters, and conferences. Signing below will give us the permission to use
the results for that purpose.

Every effort will be made to ensure that the confidentiality and privacy of participants is
protected. Your personal information will not be revealed in any reports of the results.
Only the researchers (Roy Lyster and Andrew Lee) will have access to identifiable data
(e.g., your name, date of birth, and contact information), none of which will be provided
to other study participants and be publicly disseminated. The results will be reported in
an aggregated fashion. Data will be stored in a secure hard drive. The hard drive will be
kept in a safe place locked for security purposes in the research laboratory. Please note
that the data are being kept after the study is over. This is for purely archival purposes in
keeping with university policy to keep research data for seven years following
publication.

A copy of this consent form will be provided to you. If you have any questions about this
research, are concerned about your privacy, or would like to withdraw your consent at
any time, please feel free to contact us by email at [Link]@[Link] or
[Link]@[Link].

Finally, if you have any ethical concerns or complaints about your participation in this
study, and want to speak with someone not on the research team, please contact the
McGill Ethics Manager at 514-398-6831 or [Link]@[Link].

Thank you very much.

Dr. Roy Lyster Mr. Andrew Lee

Professor Ph.D. candidate
Department of Integrated Studies in Department of Integrated Studies in
Education Education
McGill University McGill University
3700 McTavish Street 3700 McTavish Street
Montreal, QC Canada H3A 1Y2 Montreal, QC Canada H3A 1Y2
Email: [Link]@[Link] Email: [Link]@[Link]
------------------------------------------------------------------------------------------------------------
Please sign below if you have read the above information and consent to participating in
this study. Agreeing to participate in this study does not waive any of your rights or
release the researchers from their responsibilities. A copy of this consent form will be
given to you and the researcher will keep a copy.
Participant’s Name: (please print) ______________________________________

Participant’s Signature: ______________________________________________

 114

Appendix B

Instructional Materials for Form-Focused Instruction on Only Sublexical Cues

(e.g., Les Lettres Chinoises : 9)

Les lettres chinoises : 9

Je suis deux cours d’informatique le jour et un cours de français le soir.

En classe, je n’arrive pas encore à répondre au professeur, parce que très souvent

je ne comprends pas les questions. Mes réflexes semblent ralentir depuis que je

suis ici. Le professeur n’ose plus me poser de question de peur de mes « Pardon

? » À vrai dire, père, j’ai un peu honte de moi-même. Alors Nicolas m’a dit :

« Ça vient avec le temps. Pense qu’on n’a même pas, nous autres, le

courage d’aller suivre un cours en chinois ! »

Il me prête ses notes et je mets trois fois plus

de temps que les autres pour étudier les matières.

Je ne pratique pas beaucoup le français

en dehors du campus. Crois-moi, cher père, ce

n’est pas par paresse. Partout où je vais, on a

tendance à me parler en anglais. Alors,

oralement, je fais plus de progrès en anglais qu’en français. Il

est plus naturel, paraît-il, qu’un Asiatique parle anglais. Quelquefois, j’insiste

pour parler français, mais dès que je fais une faute, on passe à l’anglais. Cette

attitude m’a beaucoup découragé au début, car je croyais qu’on me parlait

anglais parce que mon français n’était pas bon. Puis Nicolas m’a dit qu’on fait

cela plutôt par courtoisie. Je découvre que la vie n’est pas vraiment
 115

insupportable pour ceux qui vivent dans un pays étranger dont ils ne maîtrisent

pas parfaitement la langue. « Ah, se dit-on, ces gens-là ont la difficulté de la

langue. » Et d’un geste las, on pardonne tout.

Je ne me pardonne pas d’avoir eu une mauvaise note pour le devoir de la

semaine dernière. Il faut que je fasse plus d’efforts. Je vous écrirai plus

longuement la prochaine fois. Je pense beaucoup à vous et à maman.

Votre fils, de Montréal

Activité 1 : répondez aux questions suivantes !

1. À qui Yuan a-t-il écrit cette lettre ?

_______________________________________________________

_______________________________________________________

2. En quelle langue Yuan suit-il ses cours ?

_______________________________________________________

_______________________________________________________

3. Pourquoi Yuan a-t-il honte de lui-même ? Trouvez trois raisons.

_______________________________________________________

_______________________________________________________

4. Qui est Nicolas ? Comment a-t-il aidé Yuan ?

_______________________________________________________

_______________________________________________________

5. (À ton avis/Selon le texte) Pourquoi est-ce que Yuan a de la difficulté à pratiquer

son français en dehors du campus ?

_______________________________________________________

_______________________________________________________
 116

6. Que suis Yuan le soir ?

_______________________________________________________

_______________________________________________________

7. Nomme trois langues que parle Yuan.

_______________________________________________________

_______________________________________________________

8. Que fait-il pour que les gens lui parlent en anglais, même s’il insiste pour parler

français ?

_______________________________________________________

_______________________________________________________

9. Qu’a-t-il eu pour le devoir de la semaine passée ?

_______________________________________________________

_______________________________________________________

Activité 2 : Masculin ou féminin ?

• En lisant le texte Les lettres chinoises : 9, remplissez les espaces

vides par LE, LA, UN, UNE, MON, MA, SON, SA, CET, CETTE, ou DU
pour indiquer si les noms sont masculins ou féminins.

• Ensuite, classifiez les noms soulignés selon leur terminaison dans le

tableau suivant le texte.

Je suis deux cours d’informatique le jour et un cours de français

_____ soir. En classe, je n’arrive pas encore à répondre au professeur,

parce que très souvent je ne comprends pas les questions. Mes réflexes

semblent ralentir depuis que je suis ici. Le professeur n’ose plus me poser
 117

de question de peur de mes « Pardon ? » A vrai dire, père, j’ai un peu

honte de moi-même. Alors Nicolas m’a dit :

« Ça vient avec le temps. Pense qu’on n’a même pas, nous autres, le

courage d’aller suivre un cours en chinois ! »

Il me prête ses notes et je mets trois fois plus de temps que les

autres pour étudier les matières.

Je ne pratique pas beaucoup _____ français en dehors du

campus. Crois-moi, cher père, ce n’est pas par paresse. Partout où je vais,

on a tendance à me parler en anglais. Alors, oralement, je fais plus de

progrès en anglais qu’en français. Il est plus naturel, paraît-il, qu’un

Asiatique parle anglais. Quelquefois, j’insiste pour parler français, mais

dès que je fais une faute, on passe à l’anglais. Cette attitude m’a beaucoup

découragé au début, car je croyais qu’on me parlait anglais parce que mon

français n’était pas bon. Puis Nicolas m’a dit qu’on fait cela plutôt par

courtoisie. Je découvre que _____ vie n’est pas vraiment insupportable

pour ceux qui vivent dans un pays étranger dont ils ne maîtrisent pas

parfaitement la langue. « Ah, se dit-on, ces gens-là ont _____ difficulté

de la langue. » Et d’un geste las, on pardonne tout.

Je ne me pardonne pas d’avoir eu une mauvaise note pour _____

devoir de la semaine dernière. Il faut que je fasse plus d’efforts. Je vous

écrirai plus longuement la prochaine fois. Je pense beaucoup à vous et à

maman.

Votre fils, de Montréal

 118

Terminaisons : Noms retrouvés dans le texte Les lettres M

chinoises : 9. Inscrivez une seule fois chaque ou
nom précédé du bon déterminant. F?
-ais, -ait, -et, -ès

-our, -oir, -ort,

-ir

-ée, -té

-tion, -sion

-ie

-esse, -isse, -asse

-ance, -ence

Activité 3 : Masculin ou féminin ?

Les mots suivants se trouvent également dans La lettre chinoise :

9, mais vous ne pouvez pas déterminer dans le texte s’ils sont
masculins ou féminins. Encerclez la bonne réponse.

1. les questions : la question / le question

2. français : la français / le français

3. courtoisie : la courtoisie / le courtoisie

 119

Activité 4 : découvrez de nouveaux mots

• Selon les régularités présentées dans le tableau de

l’Activité 2, classifiez les mots suivants selon leur genre
grammatical (M ou F ?) :

liberté, lotion, photographie, cuisse,

réservoir, finance, substance, détour
 120

Activité 5 : Une courte rédaction

Racontez une bonne expérience et une mauvaise expérience quand vous avez
essayé de parler en français ici à Montréal (5-8 phrases). Soulignez les
terminaisons qui se trouvent dans votre texte et qui se trouvent également
dans le tableau de l’Activité 2. De plus, soulignez les mots LE, LA, UN, et UNE
pour indiquer si le mot est masculin ou féminin.
_____________________________________________________

_____________________________________________________

_____________________________________________________

_____________________________________________________

_____________________________________________________

_____________________________________________________

_____________________________________________________
 121

Appendix C

Instructional Materials for Articulation-Based Instruction (Explicit Phonetic Information)

Activité 1 : Comment prononcer « un », « une », « le » et « la » ?

Un /œ̃/
Caractéristiques :
• un son nasal (l’air sort par le nez et par la bouche)
• Pincez votre nez en prononçant « un » et sentez les
vibrations !
• les lèvres arrondies
• la langue au milieu, avancée

Une /yn/
Caractéristiques :
• les lèvres extrêmement arrondies, lèvre supérieure pointée
vers le nez
• la langue vers le haut, avancée
• Les lèvres sont arrondies et projetées fortement vers l’avant
; la pointe de la langue touche les incisives inférieures
comme pour le son /i/ !
 122

Le /lə/
Caractéristiques :
• un son paresseux (un effort minimal)
• la mâchoire lâche, les lèvres arrondies très peu
• la langue au milieu, un peu avancée

La /la/
Caractéristiques :
• la bouche grande ouverte, lèvres légèrement tirées
• la langue vers le bas, avancée
 123

Appendix D

Instructional Materials for Articulation-Based Instruction

(Notice-Articulatory-Gestures and Mirror Activities)

Activité 1 : mon visage avec « un », « une », « le » et « la »

1. Choisissez le déterminant qui correspond à l’articulation parmi les deux options

données. Écrivez la réponse sous la photo.

un ou une un ou une le ou la le ou la

un ou une un ou une le ou la le ou la
 124

2. Observez les illustrations modèles de l’activité

précédente et prononcez « un », « une », « le »
et « la ».

3. Utilisez le miroir remis afin de recréer les

articulations des modèles. Au besoin, utilisez votre
téléphone cellulaire.